AVC measurements?

dirkwright · Post by **dirkwright** » Sun Jul 26, 2015 3:48 pm

Hello, I can't find complete measurements for these devices. I saw the thread about impedance and frequency response, but I need distortion and signal to noise measurements also. I need to know the total DC resistance of the AVC as well. I would like to use these in my product. I sent an email but have not heard back in several days. Anyone have any of these measurements? Thanks.

dirkwright · Post by **dirkwright** » Mon Jul 27, 2015 9:14 pm

So, no one has made measurements on these AVC's? I'm sure they are super great and I'm very interested, but I have to have some data on them before I decide. Thanks.

dirkwright · Post by **dirkwright** » Mon Jul 27, 2015 10:23 pm

I can't read private messages until I make 5 posts...

dirkwright · Post by **dirkwright** » Mon Jul 27, 2015 11:15 pm

and so it goes...

dirkwright · Post by **dirkwright** » Mon Jul 27, 2015 11:28 pm

I'm sure this is a great AVC!

dirkwright · Post by **dirkwright** » Wed Aug 05, 2015 5:00 pm

Well, simulations in SPICE are showing the typical high distortion at low frequencies, which is common with magnetic devices like transformers, chokes, etc. There is a way to mitigate that, but I haven't figured out how to apply it to an AVC yet. The THD @ 20Hz is something like 100 times that @ 1kHz.... I'm not sure I can sell a product with that much difference. Humans are not very sensitive (actually virtually deaf) to low frequency distortion, so I doubt this would have much sonic impact, but I'm afraid of reviewers laughing at my product for the large amount of bass distortion if I were to use an AVC in it. I need to think about this some more...

BTW, how much DC can these handle with the standard gap?

Post by **dave slagle** » Fri Aug 07, 2015 2:11 pm

I measure less than .1% distortion midband and with 1V input @ 20hz I get .2%. Upping the voltage to 7V @ 20hz takes the distortion to 1%. (the 1% number comes from loading a 12B4 so there may be a small bit of tube distortion. I'll post a FFT later.

No DC offset is permissable.

dave

dirkwright · Post by **dirkwright** » Fri Aug 07, 2015 3:17 pm

OK, great, thank you. My simulations show lower THD, but I don't have a good AVC model.

These numbers are with your normal air gap? What was the source impedance and the load impedance? (sounds like the load impedance was the grid of a vacuum tube?)

If the source impedance were near zero, and the load impedance near infinity, and the lams were stacked to make maximum inductance, I think that we'd see much lower THD. Do you agree?

dirkwright · Post by **dirkwright** » Fri Aug 07, 2015 6:39 pm

It seems to me that 80 H is the minimum needed to give an input impedance of 10k ohms @ 20 Hz. Expecting sources to drive something lower than that isn't wise from my perspective as a manufacturer.

Post by **dave slagle** » Sat Aug 08, 2015 3:21 pm

50R source and the load was the grid of a tube. Inductance was in the 100hy range. Tube was a 12B4 loaded with 60hy of inductance.

I think the bulk of the distortion was from the AC levels on the autoformer. They are intended for line level use so 2V or so input. 4V @ 20hz gives about 1% distortion at 20hz. The only info at 20hz in music is subharmonics and I have yet to see full signal output at that frequency.

dave

dirkwright · Post by **dirkwright** » Sat Aug 08, 2015 11:34 pm

OK, great Dave, thank you very much.

dirkwright · Post by **dirkwright** » Sun Aug 09, 2015 8:57 pm

It's entirely possible that the tube contributed a lot of the distortion you are reading. I really want to believe that your AVC has much lower distortion than what you found. I wish I could get my hands on one of these so I can connect it up to my FFT analyzer.

dirkwright · Post by **dirkwright** » Tue Aug 11, 2015 2:21 pm

FYI, from Sowter in the UK:

"9335 uses a 100% Mumetal core and the distortion over the critical range of 1kHz to 10 kHz will be less than 0.002%.

Above 10 kHz to 100 kHz and below 1kHz to 100 Hz it will rise to 0.01%.

Below 100 Hz the distortion depends very much on the applied level rising to 0.1% at +10 dBu and in a worst case scenario of +20 dBu at 20 Hz it will rise to 1%."

I wonder why a transformer has lower THD than an autotransformer?

Post by **dave slagle** » Tue Aug 11, 2015 3:17 pm

measuring 0.02% distortion is no easy task since coming up with a source that has an order of magnitude less distortion is very difficult. I was simply measuring the "worst case" situation at 20hz when I got 1% and my numbers were not far off from sowthers. (+20dBu is about 7.5V) The cause of this dostortion is simply a function of the intensity of the magnetic field in relation to saturation. More turns will lower the B field and the distortion at the cost of high frequency behavior.

dave

dirkwright · Post by **dirkwright** » Wed Aug 19, 2015 3:18 pm

Thanks. I can measure pretty good, with FFT's down to -130dB at least. [/img]

dirkwright · Post by **dirkwright** » Wed Aug 19, 2015 3:21 pm

This is a loop back on my UH-7000 line in, calibrated to -20dBFS. Noise is extremely low. It's not really necessary to measure at input voltages above 2Vrms, since that's maximum output for a CD/DAC.

Post by **dave slagle** » Sat Aug 22, 2015 2:06 pm

for reference here are some pics of my loop of my soundcard at 25hz and 1Khz. I went with 25hz to avoid the 60hz line frequency and it looks like the source harmonics are below -90dB. The 0dBfs reference is 2Vrms.

dirkwright · Post by **dirkwright** » Sat Aug 22, 2015 2:11 pm

That's good, thanks. I like that software.

Post by **dave slagle** » Sat Aug 22, 2015 2:16 pm

These all have a 2Vrms input signal and the output is taken from a -10dB tap on an autoformer and it looks like the third is -75dB at 25hz... not sure where the 2nd is coming from since an unbiased transformer should only create odd order distortions.

I'm also unclear why in the 1Khz case the loopback thrid is at -95dB and the autoformer is -105dB from the fundamental.

dave

dirkwright · Post by **dirkwright** » Sat Aug 22, 2015 3:41 pm

OK! great! That's low enough that I think I want to use these in my headphone amp product. I need to order a pair of them for testing first. How do you want to be paid?

Post by **dave slagle** » Sat Aug 22, 2015 5:02 pm

Where in the headphone amp are you planning on using them?

dave

dirkwright · Post by **dirkwright** » Sat Aug 22, 2015 6:27 pm

dave slagle wrote:Where in the headphone amp are you planning on using them?

dave

At the input, after a blocking capacitor, just like a normal volume control. The AVC is DC coupled to the + input of an opamp (probably a DOA).

You can send me an invoice via Paypal if you want to.

dirkwright · Post by **dirkwright** » Tue Oct 13, 2015 12:05 am

Hello Dave, sorry for the delay but I have been very busy lately. Here is one of several screen shots of FFT's of your AVC with a unity gain op amp buffer on the output. The performance is very good. I am impressed enough that I think I want to use these in one of my products.

Input was set at 2Vrms for 0dB. The oscillator output was then set to -10dB. I think the tap I'm using is -6dB.

Performance may improve with max inductance lam stack.

dirkwright · Post by **dirkwright** » Tue Oct 13, 2015 7:36 pm

Here is the FR, at -18dB attenuation. Note the scale. The AVC is down only 1.5dB or so at 10Hz. That's really great!

dirkwright · Post by **dirkwright** » Wed Oct 14, 2015 12:33 pm

OK, here is the FFT for -3dB input level (0dB is 2Vrms, so -3dB should be 1Vrms), with attenuation set to mid level, -18dB. Performance is still excellent. Noise is extremely low, as expected. I like this a lot.

What seems to happen is that as the level goes up, the harmonics multiply, but that could be the op amp, since it's only operating off of +/-9VDC batteries.

dirkwright · Post by **dirkwright** » Wed Oct 14, 2015 12:39 pm

I did a loop back on the L (red) channel in this screen shot. As you can see, the oscillator adds 2nd and 3rd harmonics, so we subtract that from the blue trace and see that the 2nd and 3rd added by the AVC is actually lower than shown. I can't figure out yet how to subtract the left from the right channel in the software.

You can also see that the setup of AVC + op amp adds less than 10dB to the noise floor.

dirkwright · Post by **dirkwright** » Wed Oct 14, 2015 5:04 pm

Any ideas on the temperature coefficient of resistance for this AVC? What about the voltage coefficient of resistance? In other words, how much does the resistance change with changes in temperature or applied voltage?

dirkwright · Post by **dirkwright** » Wed Oct 14, 2015 5:11 pm

The TCR of copper is 3.9x10^-3, which is 3900ppm. That's terrible for a resistor, but may not be important because this is an AVC.

Post by **dave slagle** » Thu Oct 15, 2015 1:48 pm

In this case I don't think the TCR really matters since the DCR appears as a small series resistance and when you compare the 60 or so ohms of DCR to the impedances involved a 10% change isn't going to matter.

Now when you talk about copper resistors, the TCR is a fair bit more of a concern but for some reason they still sound good (and I hate resistors). Would I use them in a situation where precise resistive value is critical like an RIAA filter.... no. (at least not yet) However there are lots of places where 10% here or there isn't an issue and when you run DC through them they tend to heat up and that often swamps any minor ambient changes.

dave

dirkwright · Post by **dirkwright** » Thu Oct 15, 2015 2:59 pm

Yeah, I don't think the TCR of the copper windings is critical to an AVC. I would be curious about the VCR though, which is the change in resistance with changes in applied voltage. A typical metal film resistor is 5ppm, for example, which is very low. So, for an AVC, the relevant value would be a change in attenuation with changes in voltage. Since we can't apply DC to an AVC, then I guess what I can do is change the AC voltage input and plot it against attenuation (voltage output). It should be a straight line. If not, then there's a problem.