|
Post by straylight on May 24, 2018 17:55:28 GMT -5
Ok so I want to get some usable data on the pickups I'm building, why they're good, why they're awful. I've started down the Ken Willmott route using a sound card and Rightmark. I have some boards for an integrator circuit in the post. I've made a driver coil by winding 300 turns of 0.212mm polysol magnet wire onto a spare humbucker bobbin, thrown a 100Ohm resistor in series, soldered it to a short (6") patch cable, wrapped in in loom tape and potted it in beeswax/parafin wax blend. As an aside I think I shall be dropping the beeswax as I'm beginning to see some green on humbucker baseplates. I've put crocodile clips and short fly leads on the end of another short patch cable. I've got an M-Audio Delta series soundcard surplus from the studio. There's not a lot in the way of level controls easily available with the windows drivers, so a 3m RCA/jack throws it at a small behringer eurorack. Sound card goes into a mono channel with unity gain and sent out to AUX 1, which my driver coil plugs straight into. My crocodile clip lead plugs into a behringer DI which returns to a second channel on the board I adjust the preamp here for rightmark (or anything else) to be happy about levels and bounce straight out to AUX 2 which returns to the sound card. It's a simplified version of how i've recorded with a line level mutli-channel sound card for two decades with reliable results. Rightmark shows no anomalies if I replace the driver coil and pickup with a patch cable. I've made a simple 200kOhm 470pF dummy load as appears standard to represent guitar controls and a cable. Then this notch that I now can't reproduce appeared and made a mockery of me. Note to self: verify everything that looks interesting. So that's not helpful. I've got a few other bits I've ran tests for and it's conveniently sabotaging me by normalising the height of the spectra, is it perhaps constant area under the curve? This inductive notch filter on a strat style SC
My next stage is to try and get good figures using R and the seewave library, i'm not particularly competent with it, but it's been a decade since i had a matlab license. This means generating a noise burst and a sweep as a wave file, bouncing it through the board and recording the wave produced, and then batch processing the results. Annoyingly I'm getting a lot of noise somewhere that rightmark doesn't see. I'm wondering if it's aliasing noise from sampling at less than 10x the frequency. The obvious answer here is my cheap audio kit is not up to the task. I could wrestle a better interface out of the studio but that's a lot of hassle. I'm intrigued by the syscompdesign and vellman USB function generators and scopes. They're kind of justifiable as I've been using a very old Hitachi oscilloscope to troubleshoot amps, and my cheap signal generator is a weak link. Does anyone have significant experience with either?specifically i want to know if I can dump data sampled when doing a frequency response plot to fine for external processing.
|
|
|
Post by stratotarts on May 25, 2018 7:57:10 GMT -5
There are multiple facets to getting something like this working, as the end result is a mix of factors and it's hard to tell what part might be causing a problem if there is one. Also I'm not familiar with the equipment you have there, perhaps photos would be helpful. I worry about the output impedance of the "small behringer eurorack". A line level audio out may have an impedance as high as 2,000 ohms and is not low enough to drive the coil properly, also the output level may not be high enough for all tests. In contrast, a headphone output has no more than about 200 ohms impedance, a much higher potential drive level, and usually has a convenient way of adjusting the level. You can not get absolute level readings with Rightmark, it's a quirk of the software. For that you need to use a software oscilloscope. It is not necessary to wax pot the drive coil. Your physical component layout may need work, I see a lot of wiggle in your readings, I got much smoother results like this:
Cheapness certainly isn't an issue, as I have had people report success with the Behringer UC222, maybe the cheapest ADC out there. It's my impression that you have far too much hardware in the audio chain. I tried using a Mackie mixing board myself, and it was a failure, the readings were all over the place.
|
|
|
Post by straylight on May 27, 2018 20:55:48 GMT -5
It's a venerable MX1604. It's handy for miking up guitar amps and generally abusing. I've got access to better, but not that I can depend on to be available on my desk every time I want it. Spec: media.music-group.com/media/PLM/data/docs/P0087/MX1604A_P0087_S_EN.pdf I could try using the headphone out and see if it makes a difference. I've been running line level soundcards with RCA or 1/4" sockets into patchbays and racks of kit for decades. There's probably a few Creative cards kicking around with mic and headphone sockets. Most of which I've superglued back together a few times as they just don't stand up to heavy use. It's probably the cheap way to remove kit from the signal chain, but compared to typical recording applications my signal chain is really short. I think this just strengthens the argument for a USB oscilloscope. I've sold enough pickups in the last week to probably justify it.
|
|
|
Post by stratotarts on May 29, 2018 9:59:12 GMT -5
In your case, a USB oscilloscope would be a good choice. It is hard to visualize a block diagram from your description, but it sounds like it has way too much equipment in it. You mentioned a Behringer DI, that would normally be all you would need for this. The link you provided was useful, but more links or specific equipment ID's and clear photos of your actual experimental setup are necessary in order for anyone to help you, if you want to proceed in the direction of using Rightmark.
|
|
|
Post by antigua on May 29, 2018 12:14:23 GMT -5
The one thing that bothered me about the Rightmark setup was that it wouldn't let you see the resulting bode plot until after it ran through the entire sweep, where as the USB oscilloscopes I've used, Velleman and SysComp, show the plot in real time, so that if there is an error, it's immediately obvious and you don't have to way 60 second or whatever it might take.
|
|
|
Post by straylight on May 30, 2018 21:37:41 GMT -5
I have discovered that the level and peak lights on my board don't come anyway near triggering if I've got a sine wave distorting. Found out what was going on using JAAA (a not particularly useful realtime spectrum analyser for linux/jack) and listening to the signal going to the driver coil and coming back from the DI and watching for harmonics appearing. Looks like the behringer preamps will take quite a beating at 1kHz without clipping, but there's not as much headroom at 7kHz.
I've sold or have orders for every pickup I have flatwork for over the course of the bank holiday weekend so I'm definitely going down the usb scope route. I have been investigating the vellman and syscomp products. I like the way the vellman software runs in demo mode if you don't connect a device.
|
|
|
Post by stratotarts on May 31, 2018 8:18:12 GMT -5
As I said, a USB would be good for you. However, it is easy to detect and troubleshoot distortion issues with Rightmark/DI. Although the sweep is not active, the real time spectrum analyzer is. So all aspects of signal quality, level, noise, mains interference, and harmonic distortion are plainly visible on the display screen when the pilot tone is active (which is always is, prior to running a sweep). It only shows the conditions at the pilot frequency (configurable, 1 kHz) but that is sufficient to establish acceptable signal quality, except in highly unusual circumstances. This is documented in my guide to using the integrator.
If you have preamp clipping at 7 kHz, your drive level is too high.
|
|
|
Post by straylight on Jun 1, 2018 18:26:18 GMT -5
I'm getting what I think are consistent and repeatable results out of my setup, I need to run it between -3dB and -6dB on rightmarks 1khz pilot tone with everything on the board set to unity gain except the preamp gain after the DI.
I have, however, caved and ordered a SysComp 101 which should have adequate range and plenty of oversampling for anything audio and i'm not doing anything RF. For a start it's comparable in price to an inexpensive proper signal generator and the java and pocket money electronics kit versions aren't giving stable enough waves to properly setup a tube amp.
I'm just waiting for some diodes that should have arrived yesterday and I'll be able to assemble the integrator.
It's kind of frustrating that the only pickup I have that anyone else has posted full stats on appears to be the Air Norton. I guess when I'm confident everything will be getting posted and will be useful.
I have a cheap LCR meter that seems to be measuring at 55Hz in the 20H range according to my analog hitachi scope. How trustworthy are the inductance readings likely to be? Checking against the capacitance swamping method there's maybe 10% difference. And it's not coupling with the driver coil, measuring a pickup with the driver coil taped on doesn't even appear to affect the 3rd (and last) significant figure.
P-bass coils. I've got a whole heap of them. It's how I got into this in the first place. How do I go about arranging the driver coil and pickup halves for consistent readings? In a p-bass pickguard with the driver between them where they meet or pickup coils arranged like a regular humucker?
|
|
|
Post by stratotarts on Jun 6, 2018 13:50:02 GMT -5
If you can not get it to stabilize at -2dB or so, there is still some loop gain issue. The problem might go away when you drive it with the sig gen instead.
The Extech has been established as more or less the gold standard for pickup inductance measurements. I would worry about mains interference if measuring at 55Hz. But it all depends on the actual meter circuit and calibration. The nice thing about the Extech is that it is usually the one that was used for published inductances. You should set it to measure at 100 Hz.
Exciter coil placement for the P-bass would be "whatever you can get away with". I would just treat it like a single coil and place the coil inline with one of the pair.
|
|
|
Post by straylight on Jun 13, 2018 22:16:30 GMT -5
Ok, integrator is together, waiting for a project box in the post but it *is* working.
Small eurorack has a number of irritating crackles in sealed switches and needs a major soldering project or errm, well, retiring.
Found incredibly inexpensive Logitech usb audio adaptor amongst the collection of misc cables, output's a bit high and it only has software volume controls so I had to get creative, but results are consistent, still waiting for syscomp in the post.
55Hz induction test is showing some sensitivity to the orientation of pickups of maybe +/- 0.01H, is consistent with integrator produced readings. Readings from integrator fed back into a spice model are closer than expected.
What I can't work out (yet) is is a mathematical derivation of the Q factor for an idealised pickups across a test load, or whether indeed there's so much of the test conditions in calculating such that I may as well read of the plots.
|
|
|
Post by stratotarts on Jun 16, 2018 23:04:22 GMT -5
What I can't work out (yet) is is a mathematical derivation of the Q factor for an idealised pickups across a test load, or whether indeed there's so much of the test conditions in calculating such that I may as well read of the plots. Ignoring the coil resistance, it's Q = R L*(C/L) 0.5
|
|
|
Post by straylight on Jun 17, 2018 1:17:11 GMT -5
Thanks, I did have the right formula, I was just feeding it the wrong R because failure to think.
|
|
|
Post by stratotarts on Jun 19, 2018 20:15:48 GMT -5
The formula I have for the loaded Q including coil resistance is: QL = (L*CL*(1+R/RL))0.5/(R*CL+L/R)
|
|
|
Post by straylight on Jun 20, 2018 17:31:28 GMT -5
Thanks guys, you are being really helpful. I'm feeling kinda awkward holding back on my data but right now I'm not getting good data and I don't want to be caught in a loop of publish, retract, repeat. I've studied and worked in MechEng and CompSci and it's always a wake up call when I stick my nose into a different field out of necessity and the complexity of a problem unfolds.
I've got an Air Norton and a SuperD on my becnch to compare my data, most of what I have is more metal oriented but I am at least getting increasingly consistent qualitative results I can use to compare generations of my handwound specs.
The syscomp has finally arrived, it's an interesting little unit and the signal generator is useful for troubleshooting amps. I'm not using it to poke around amp internals though, that's for a a decent voltmeter and a standalone analog scope.
I've got a few issues though, it looks like my cheap propes are both 80pF on 10x and 140pF and 144pF on 1x. This seems high. They were very cheap, and have been set up for a nice square square wave on a 0.5v 1kHz calibration signal in the 10x range. I'm sure I can get better for sensible money, I'm seeing 20pF noted here a lot. Is it a case of look for somehting advertised in that range and test to see if it's close, or are there some affordable brands I should be looking for?
Am I correct that I don't need a super-low capacitance probe on the driver-coil side of things or have I missed something?
Cheap meter measures capacitance at 110 Hz, and I'm holding the test probes has i would for the reading, with the scope probe in the same hand and cables falling as they would if they were connected, noting the free cable capacitance, connecting things up and subtracting that value. I think that's the right thing to do.
I'm getting some "staircase" behaviour, it's happeing on every pickup and is repeatable, hapening at the same places eveyr time. Any ideas what's going on here? I hope it's not an artefact of the CGR changing ranges. It's not where the interesting bit of the curve is, but it's still somehting i need to sort out.
And checking I have understood this all correctly, stratotarsts equation for Q with inline and Load resistance is as below where the circuit is reduced to the diagram below RL and RC being the parallel sums of all capacitances and resistances (so 1M amp load, volume pot, pickup capacitance as calculated, cable capacitance as calculateted/mesured/simulated etc) yes?
Once again i've either got it or I'm horribly wrong and I'm about try stacking scope and spice data in excel. I can at least get my head around-20db/decade whilst I wait for project box and various connectors to arrive. Longer-term plans involve using R (kind of like matlab, but more statisitics orianted and open source) to automate curve matching for display and bulk data matching. What are you guys using to overlay bode plots?
|
|
|
Post by straylight on Jun 20, 2018 19:13:18 GMT -5
Yes, missed something.
Utter brain fail on probe capacitance. In the 10x range they are compensated to appear as zero capacitance at 1kHz which I just damn well set. Does this compensation include the 20pF indicated capacitance of the scope? My inclination is that it does as I'm calibrating differently with the SYSCOMP as i would be with my old Hitachi V-223. I need to find the manual for that, but I *think* it's the same as my (sadly malfunctioning) Gould at 28pF as I didint have to adjust the probes swapping between the two.
|
|
|
Post by antigua on Jun 21, 2018 1:05:48 GMT -5
Thanks guys, you are being really helpful. I'm feeling kinda awkward holding back on my data but right now I'm not getting good data and I don't want to be caught in a loop of publish, retract, repeat. I've studied and worked in MechEng and CompSci and it's always a wake up call when I stick my nose into a different field out of necessity and the complexity of a problem unfolds.
I've got an Air Norton and a SuperD on my becnch to compare my data, most of what I have is more metal oriented but I am at least getting increasingly consistent qualitative results I can use to compare generations of my handwound specs.
The syscomp has finally arrived, it's an interesting little unit and the signal generator is useful for troubleshooting amps. I'm not using it to poke around amp internals though, that's for a a decent voltmeter and a standalone analog scope.
I've got a few issues though, it looks like my cheap propes are both 80pF on 10x and 140pF and 144pF on 1x. This seems high. They were very cheap, and have been set up for a nice square square wave on a 0.5v 1kHz calibration signal in the 10x range. I'm sure I can get better for sensible money, I'm seeing 20pF noted here a lot. Is it a case of look for somehting advertised in that range and test to see if it's close, or are there some affordable brands I should be looking for?
Am I correct that I don't need a super-low capacitance probe on the driver-coil side of things or have I missed something?
Cheap meter measures capacitance at 110 Hz, and I'm holding the test probes has i would for the reading, with the scope probe in the same hand and cables falling as they would if they were connected, noting the free cable capacitance, connecting things up and subtracting that value. I think that's the right thing to do.
I'm getting some "staircase" behaviour, it's happeing on every pickup and is repeatable, hapening at the same places eveyr time. Any ideas what's going on here? I hope it's not an artefact of the CGR changing ranges. It's not where the interesting bit of the curve is, but it's still somehting i need to sort out.
...
What are you guys using to overlay bode plots?
I have probes like these www.amazon.com/Signstek-2PCS-P6100-Oscilloscope-Probe/dp/B00ENF272C/ref=sr_1_1?ie=UTF8&qid=1529560322&sr=8-1&keywords=oscilliscope+probe , they spec 18pF at 10x and 90pF at 1x, so it almost sounds like you're getting 1x performance. You can also get a lot of capacitance from shielded lead wires of humbuckers, in the area of 70pF for a one foot lead, but I'm not sure if that's an issue in your testing. If you're using the Ken Willmott integrator, the capacitance shouldn't matter much in that case either, because the probe connect to the integrator (or directly via a BNC connector), and then the issue is the capacitance between the pickup and the integrator, which I worked out to be about 10pF with these short little wires that are attached to it. At this point you have to actually make sure the leads and pickup lead wires are as far apart as possible, and if it's a shielded lead, you may or may not want to account for that large, unavoidable capacitance in your survey of data. I just include it, since it's always going to go hand in hand with the pickup itself. So that's another thing to watch out for: if you have a humbucker with a four lead pickup wire, make note of whether the shielding is or is not included in the plot that is made. I'd suggest including since it would be connected in situ. It's true that the capacitance on the driver side doesn't need to be low, though we're assuming that even with a high capacitance it's own resonant frequency is very high, like ten times higher than the highest test freq uency, so that it's own output is flat.
The wiggly staircase is something I got with the SysComp also, I don't think I ever figured out what it was. If you decrease the size of the frequency steps it will suppress that somewhat. I don't have such issues with the Velleman PCSU200 or PCSGU250. I think ultimately it's a hardware issue. The Q factor math, I'm not so sure about either. For the overlaying of plots, the Vellemans support up to 5 overlays per plot, or more, but then they're just black traces from the 6th onwards. The SysComp you have should support two traces, the first trace turns light gray, which it looks like you've discovered. The Rightmark software supports multiple successive plots also, I don't remember exactly how many, though.
|
|
|
Post by straylight on Jun 21, 2018 2:49:04 GMT -5
The wiggly staircase is something I got with the SysComp also, I don't think I ever figured out what it was. If you decrease the size of the frequency steps it will suppress that somewhat. I don't have such issues with the Velleman PCSU200 or PCSGU250. I think ultimately it's a hardware issue.
Well that's a bit annoying. Especially as the Syscomp website features an article on using an older product in the same range to perfome just this task.
|
|
|
Post by stratotarts on Jun 21, 2018 12:35:23 GMT -5
Yes, missed something.
Utter brain fail on probe capacitance. In the 10x range they are compensated to appear as zero capacitance at 1kHz which I just damn well set. Does this compensation include the 20pF indicated capacitance of the scope? My inclination is that it does as I'm calibrating differently with the SYSCOMP as i would be with my old Hitachi V-223. I need to find the manual for that, but I *think* it's the same as my (sadly malfunctioning) Gould at 28pF as I didint have to adjust the probes swapping between the two. I assume you are talking about the screw adjustment on the probe. That compensation is not for capacitance. It's for frequency response. No matter how you adjust it, it will still present almost the same capacitive load at the probe tip. The capacitive load of 10x probes is surprisingly high on average. The way around that is to use a 100x probe. Those have much lower capacitance. With a probe in 10x or 100x mode, the scope input capacitance is not really very important because it's swamped by the probes internal parallel compensation capacitance.
|
|
|
Post by stratotarts on Jun 21, 2018 12:38:45 GMT -5
[...] Cheap meter measures capacitance at 110 Hz [...]
That sounds way too low for measuring picofarad values. 90pF is more than you would expect from even the cheapest 10x probe.
|
|
|
Post by stratotarts on Jun 21, 2018 13:55:43 GMT -5
And checking I have understood this all correctly, stratotarsts equation for Q with inline and Load resistance is as below where the circuit is reduced to the diagram below RL and RC being the parallel sums of all capacitances and resistances (so 1M amp load, volume pot, pickup capacitance as calculated, cable capacitance as calculateted/mesured/simulated etc) yes?
Yes, except I think you mean C L, not RC. The problem is, I lost the derivation for this a long time ago. I'm too long away from the original work to remember how I came up with it. I tried referencing some articles that I thought I got it from, but no luck. However, it is a column in the spreadsheet that I use for all my test results. Because 20*log 10(Q) gives you the relative amplitude of the loaded resonant peak, I have been able to match it with a lot of measured results. It has always come very close for pickups without expected large eddy losses, such as Alnico single coils.
|
|
|
Post by straylight on Jun 22, 2018 0:53:17 GMT -5
Thanks.
Emailed syscomp support, had reply from Peter Hiscocks so at least the problem domain is known, i'm not sure how big syscomp is though, might be something that can be sorted. I wonder if I get the same results driving the scope with serial commands....
|
|
|
Post by straylight on Jun 27, 2018 17:09:39 GMT -5
Well there's new a netalyser.tcl for the Syscomp scope on my system and it is in the process of being rolled out into a new package on their website as we speak and it's working. It's plotting a visually smooth line, gonna have a prod at the .csv output and see just how good, but I'm really impressed with how quickly they sorted the issue and got new code out.
|
|
|
Post by antigua on Jun 27, 2018 23:58:58 GMT -5
Well there's new a netalyser.tcl for the Syscomp scope on my system and it is in the process of being rolled out into a new package on their website as we speak and it's working. It's plotting a visually smooth line, gonna have a prod at the .csv output and see just how good, but I'm really impressed with how quickly they sorted the issue and got new code out. Thanks for the update. I'll give the CGR-101 another try.
|
|
|
Post by straylight on Jun 28, 2018 8:13:11 GMT -5
IIRC the CGR-101 was 20MS/s input and I don't know the resolution of the signal generator was.
The CGM-101 still runs at well over audio frequencies and has ample oversampling for the task, looking at the 200kHz signal on my 20MHz analog scope there's some visible aliasing on the wave but at 1kHz and 10kHz there's a very stable and smooth wave. Interestingly i'm seeing a slight saw tooth to it (rising faster than falling). I think this is an issue with the hitachi as it's not apparent on my Gould, but the Gould has an X-sweep issue at the moment and I've only got about 1/3 of the screen to play with.
One thing to be aware of with the SYSCOMP kit is it's all ground coupled to the PC host. This makes me a bit nervous in some applications. I think i'm safe if my PC is floating (so laptop on battery) but as I've had an amp malfunction and grounding issue take out a connected laptop and USB soundcard i'm still a bit paranoid doing amp repairs, off topic I know but if anyone can point me at some practical resources for not frying my test equipment doing amp repairs that'd be handy.
Ok, verifying my probe performance: I'm fairly sure I'm not getting a correct reading using an LCR meter after having looked at a circuit diagram for the probes, but does my logic follow here: 1x operation should be without an series capacitance and I can measure that with the meter consistently. If I then measure a series of pickups with probes at 1x and 10x and note the recorded capacitance, the difference in capacitance should be the difference between 1x and 10x operation yes? This gives me 20pF (+/-1pF ) very consistently so it looks like probes are to spec. I'm getting sane results from my pickups as well.
|
|
|
Post by stratotarts on Jun 28, 2018 23:01:47 GMT -5
One thing to be aware of with the SYSCOMP kit is it's all ground coupled to the PC host. This makes me a bit nervous in some applications. I think i'm safe if my PC is floating (so laptop on battery) but as I've had an amp malfunction and grounding issue take out a connected laptop and USB soundcard i'm still a bit paranoid doing amp repairs, off topic I know but if anyone can point me at some practical resources for not frying my test equipment doing amp repairs that'd be handy.
Ok, verifying my probe performance: I'm fairly sure I'm not getting a correct reading using an LCR meter after having looked at a circuit diagram for the probes, but does my logic follow here: 1x operation should be without an series capacitance and I can measure that with the meter consistently. If I then measure a series of pickups with probes at 1x and 10x and note the recorded capacitance, the difference in capacitance should be the difference between 1x and 10x operation yes? This gives me 20pF (+/-1pF ) very consistently so it looks like probes are to spec. I'm getting sane results from my pickups as well.
If you only need one channel, the time honoured method of "floating ground" is to connect the hot lead of 2 different channels, and enable "sum" and "subtract" mode. That gives you a floating differential reading. You are fine as long as you don't exceed the rated voltage of the inputs.
I'm not sure what you are saying about the capacitance. Why are you considering 1x measurement or 1x capacitance at all? I don't see the purpose. Looking at the circuit diagram for the probes won't give you an accurate idea of the actual capacitance because the stray capacitance due to the conductor layout inside it is significant compared with the divider caps. The input capacitance in 1x mode is usually much larger than in 10x mode (that is actually the purpose of 10x mode - the lowest practical input capacitance).
|
|
|
Post by straylight on Jun 29, 2018 8:42:33 GMT -5
Two probes and subtract for floating ground is useful, thanks.
I'm not planning on using 1x for expermental records, just equipment verifcation.
I can measure the capacitance of the 1x probe with the multimeter and it behaves as expected.
10x is an RC network with series capacitor in tip and parallel variable capacitor at the BNC socket. Low freqency multimeter returns out of spec results.
I can measure a pickup (or series thereof) at 1x and solve for capacitance, I get pickup + probe at 1x (known), then measure at 10x and get pickup + probe at 10x (want). This gives me a result across 10 pickups of 20pf +/-1pf which is a) probe spec, b) less variance than probe spec, and c) surprisingly precise given the confidence I don't have in the initial meter reading of the 1x probe.
Either i've missed something in this methodology or the probes can be trusted to be spec effective capacitance of 20pf +/- 2pf at 10x which is what I was trying to verify. It's still a significant amount given pickups are O100pF, which if not using the integrator will need to be factored out of my calculations, and equally i can use data on pickups with the 10x probe to verify capacitance of the integrator. Talking of which the case finally arrived in todays post...
|
|
|
Post by stratotarts on Jun 29, 2018 9:38:11 GMT -5
I don't think you're necessarily getting a more accurate reading of the capacitance at 1x than at 10x. Whatever meter you're using probably has a large margin of error at the 100pF or so that the 1x has. I think I understand how you're using the two readings to derive the 10x, but although the math and technique is correct, I think the margin of error going into it skews the result by amplifying the error. If you want to be really confident of the input capacitance, you will have to find a way to measure the 10x input capacitance more directly, with a more accurate capacitance meter in my opinion. Is there a reason you are not confident with the manufacturer's specification of input capacitance at 10x?
Also I doubt that the assumption of 2% accuracy is realistic (2pF/100pF). Not only will it change that much any time you so much as move a wire, but it is way beyond the need for accuracy in the application scenario. I haven't done any math on this, but I'll bet that 2% difference in capacitance would probably make about 0.001 dB difference in an actual guitar circuit.
You could make an RF LC oscillator with known values of capacitance and then substitute the probe input for the C. That would get you an accurate measurement if you can't obtain a really accurate meter.
|
|
|
Post by straylight on Jun 29, 2018 13:32:49 GMT -5
Mostly I'm checking everything out of paranoia and my early test test setup has thrown up equipment glitches and anomalies everywhere. There's inaudible but measurable crosstalk and nonlinearities in my behringer RX1604 (retired rack mixer, formerly for live keyboards, now desktop monitors and headphones), and occaisonal dropouts caused by I think corroded switches in my behringer MX1604 (retired utility mixer, now for spares). I needed an update from Syscomp for the network analyser to work correctly on my CGM-101. I've identified several faulty patch cables and rebuilt my pach bay. I have a pickup that sounds awesome in my SIR70FD giving the same resonant peak as the Tone Zone which sound rubbish in the same instrument and is a large franction of why I'm winding pickups.
I have malfunctioning kit and uncontrolled variables and right now I trust nothing I can't in some way prove to be working.
I'm double-checking my frequency peaks using my Hitachi V-223 and Lissajous figures, it's time consuming but it's increasing my confidence in the data. The only items I cant control for are the probes. They're cheap ones and I can't drectly confirm the effective capacitance with the meter. I could do with a better LCR meter, something benchtop that reads more thant 3 significant figures and trustworthy to, well, any significant figures really, but I'm trying to fund guitar related activities out of sales and repairs which aren't particularly lucrative. The multimeter probes i'm using with the LCR meter are between 10pF and 40pF depending on wheter they are held far apart or in a tangle and it takes some patience to hold the meter and leads steady for a consistent unconnected reading. Today I'm getting 18pF unconncted, 55pF connected to the tip of the probe whilst the other end is plugged into my syscomp and 93pf loose. 1x comes in at 158pF loose, which corrected for my multimeter leads is 140pF and spec. I know my reading of the 1x is likely a bit fluffy, that they have come back as confidently spec is less worrying than me trying to compensate out a new value. The sane thing to do is see if I can borrow a posh or at least verified probe and compare against a pickup I've measured (hard) or compare my readings for things I have against what's published here. I think I'm about ready to say I'm confident.
|
|
|
Post by stratotarts on Jun 29, 2018 14:14:39 GMT -5
What is the difference in reading with the DVM probes connected to the scope probe, if you can hold everything still and connect and disconnect it? That should take care of stray capacitance in the DVM probes. Confused about the CGR-101. Apart from some ripples in the graphs, isn't it basically working?
|
|
|
Post by antigua on Jun 30, 2018 13:51:01 GMT -5
Mostly I'm checking everything out of paranoia and my early test test setup has thrown up equipment glitches and anomalies everywhere. There's inaudible but measurable crosstalk and nonlinearities in my behringer RX1604 (retired rack mixer, formerly for live keyboards, now desktop monitors and headphones), and occaisonal dropouts caused by I think corroded switches in my behringer MX1604 (retired utility mixer, now for spares). I needed an update from Syscomp for the network analyser to work correctly on my CGM-101. I've identified several faulty patch cables and rebuilt my pach bay. I have a pickup that sounds awesome in my SIR70FD giving the same resonant peak as the Tone Zone which sound rubbish in the same instrument and is a large franction of why I'm winding pickups.
I have malfunctioning kit and uncontrolled variables and right now I trust nothing I can't in some way prove to be working.
I'm double-checking my frequency peaks using my Hitachi V-223 and Lissajous figures, it's time consuming but it's increasing my confidence in the data. The only items I cant control for are the probes. They're cheap ones and I can't drectly confirm the effective capacitance with the meter. I could do with a better LCR meter, something benchtop that reads more thant 3 significant figures and trustworthy to, well, any significant figures really, but I'm trying to fund guitar related activities out of sales and repairs which aren't particularly lucrative. The multimeter probes i'm using with the LCR meter are between 10pF and 40pF depending on wheter they are held far apart or in a tangle and it takes some patience to hold the meter and leads steady for a consistent unconnected reading. Today I'm getting 18pF unconncted, 55pF connected to the tip of the probe whilst the other end is plugged into my syscomp and 93pf loose. 1x comes in at 158pF loose, which corrected for my multimeter leads is 140pF and spec. I know my reading of the 1x is likely a bit fluffy, that they have come back as confidently spec is less worrying than me trying to compensate out a new value. The sane thing to do is see if I can borrow a posh or at least verified probe and compare against a pickup I've measured (hard) or compare my readings for things I have against what's published here. I think I'm about ready to say I'm confident.
Have you considered trying the Velleman PCSU-200? It's not very expensive either, and it has superior bode plotting features, IMO.
|
|