Yes. You can install 470R resistors, one each between PD+ and ND- to the NFB.
But it is very bad method, maybe only for elementary control for big mistakes, but for such test you can connect +PD,-PD and NFB direct , no need for 470R... VAS output (high impedance, that is why we use predrivers to minimize VAS loading and so ensure linearity) is direct loaded with low impedance NFB network and linearity and NFB margin is than very poor .
And input coupling cap value has nothing to do with rounded square, but HF filter capacity (150-680p?) , together with generator output impedance (compare input and output signal). But this CFA should be very fast. Terry, can you post a picture of this rounded square? And here,
And input coupling cap value has nothing to do with rounded square, but HF filter capacity (150-680p?) , together with generator output impedance (compare input and output signal). But this CFA should be very fast. Terry, can you post a picture of this rounded square? And here,
you mean 884 miliohm or 0.884 megaohm? I suppose , the second value.
Last edited:
I haven't seen any difference in high frequency square wave changing value of input capacitor .
This occurs only for frequences under 50Hz
distortion measurments.
I believe that i know the reason for this increased distortion level now.
Look at picture No 4.What did you see here?
Distortion go up when frequency rises.
Why this occurs only when a load is connected?
Probably because this bad load has big electromagnetic value in those high frequencies.
I repeated all these tests without load and within a real loudspeaker.
I believe that i know the reason for this increased distortion level now.
Look at picture No 4.What did you see here?
Distortion go up when frequency rises.
Why this occurs only when a load is connected?
Probably because this bad load has big electromagnetic value in those high frequencies.
I repeated all these tests without load and within a real loudspeaker.
Attachments
Last edited:
Nice work Terry
Hi Terry,
Nice work indeed being persistent and chasing that issue down. So just to be certain I'm clear you currently have BV's latest MOD working...though one he tweaked slightly after the BV2 mod to help with clipping behavior?
If this is the case I will place an order for my IPS parts today so I can join the fun.
hopefully I can get better measurements off my HP 35665a than what a good sound card can provide.
Again awesome work Terry...you make me look forward to retirement...I'd love to play with this stuff all day long.
Hi Terry,
Nice work indeed being persistent and chasing that issue down. So just to be certain I'm clear you currently have BV's latest MOD working...though one he tweaked slightly after the BV2 mod to help with clipping behavior?
If this is the case I will place an order for my IPS parts today so I can join the fun.
hopefully I can get better measurements off my HP 35665a than what a good sound card can provide.
Again awesome work Terry...you make me look forward to retirement...I'd love to play with this stuff all day long.
Hi Guys,
I'm not sure i'd pull the trigger on the BV mod just yet. Here are some scope shots. First three are stock OS CFA with 4.7uF input. Second three are BV mod #1 4.7uF. Third three are mod BV #2 with 10uF input. 20kHz, 50kHz, 100kHz
I'm not sure i'd pull the trigger on the BV mod just yet. Here are some scope shots. First three are stock OS CFA with 4.7uF input. Second three are BV mod #1 4.7uF. Third three are mod BV #2 with 10uF input. 20kHz, 50kHz, 100kHz
Attachments
R3=10R, R2= 10k..
Thimios, here can be nothing feromagnetic in your load , e.g. resitor caps or leads (try it with permanet magnet) or near load , it will act as a core, and can saturate, result is nonlinear load resitance-distortion. The same is valid for output inductor in amp, worst thing what one can do is to wound it on paraleled resitor. I hope, it is not this case..
Terry, if you want "sharper" square, you need to decrease input filter capacity (680p to some 150-220p, as in slewmaster), input signal (generator source impedance loaded with this capacity) looks about the same as amp output. I posted picture about this few days ago..
Last edited:
Finally a new signal generator...
This is what a 100kHz square looks like with my unmodified OS CFA-XH operating in isolation on +/-45V, input filter in place set to ~1.5MHz cut-off (220R-470pF). I'm tempted to pull C2 just to see how an un-filtered IPS looks.
This is what a 100kHz square looks like with my unmodified OS CFA-XH operating in isolation on +/-45V, input filter in place set to ~1.5MHz cut-off (220R-470pF). I'm tempted to pull C2 just to see how an un-filtered IPS looks.
Attachments
The high frequency distortion is still really bad. I think it's more reasonable to assume that the rise perhaps doesn't occur until later because the three ways impedance might be quite higher throughout the mid band, this would cause less current to flow inside the amplifier and reduce the distortion.
Spooky with no load should measure as your sound card measures. Spooky should have 2nd order probably around 0.0005% or so across most of the audio band, only rising with the fall in negative feedback at higher frequencies. As far as I am aware the 3rd order harmonic with spooky should be lower than the 2nd, but as your sound card has gross 3rd order here you're not going to be able to measure it accurately.
If your sound card reliably measures and produces very low levels of 2nd harmonic at various drive and measurement levels, then I would expect the no load measurement to pretty much look exactly like the sound card does.
The loaded measurement, at least into 8 ohm loads at moderate power levels, should look pretty much like the unloaded measurement. Low frequency distortion levels of <0.0005% with this only rising when feedback starts to decrease due to cdom. If you're getting a rise in distortion before cdom starts to operate then this is being caused by something other than the amplifier circuitry.
You've still got contamination occurring somewhere. Either induced within the amplifiers layout or due to the measurement configuration.
You're obviously not getting a traditional ground loop here as the FFT is completely clean, but I do think you have high level current flowing somewhere that you don't want it to, like down the shield to the input wire.
This is getting complicated
I only just realised that in the loopback measurement with ASIO 2nd order is dominant. In the loopback without ASIO the 3rd order is dominant.
Have you tried doing what I suggested before with the windows settings to try and get your card to actually work at 96 or 192kHz?
We can see that it isn't actually working beyond 48k because of where the distortion plots stop. When you're in 44.1/48k you get the 3rd HD up to around 8.5k and the 2ndHD stops at around 10k. If you manage to get the system to operate at 96k both these frequencies double.
Hey 5th, nice looking board. Looking forward to your review. I gave all three CFA boards a listen to today. They sound identical. I guess if you're into measuring you may find one superior to another but for me, the stock version sounds every bit as good as the modded versions. As as matter of fact, you see my screen shots. The latest version has the worst looking 100kHz waves. Of course, BV did say that layout would be important and I was pretty much limited to what I had. Hopefully your new layout will pan out better.
A pic from someone who has too much time on his hands.
A pic from someone who has too much time on his hands.
