New Doug Self pre-amp design...

[DouglasSelf]

Since alarm and despondency over the performance of the active volume control at very low settings seems to be spreading, I thought it was time to put the speculation on hold and look at some facts.

The version I have to hand uses a 5K pot and LM4562 dual inverters. Assume the control is calibrated 0 - 10.
Measured 20 Hz - 50 kHz.

Test 1: Volume set to 1. 20 Vms in, and 1.6 Vrms out. THD residual indistinguishable from the GenMon output of the AP SYS-2702

Test 2: Volume set to 0.2 (ie 2% off endstop) 20 Vms in, and 300 mVrms out. THD residual is entirely noise with no trace of non-linearity.

I then replaced the 4562 in the dual inverters with a Texas 5532.

The results are very much the same except the noise level in Test 2 is perceptibly higher at 0.0008% rather than 0.0007%.

I suggest that we do not have a problem here.

[owdeo]

Thank you for doing this Mr Self. I was really hoping someone would be in a position to, and who better!

In your published work over the years your position on the merit of assessing performance subjectively has been crystal clear. I can offer no scientific proof that there is anything wrong with the sound qulaity and absolutely appreciate that you would think that this effect I'm hearing isn't real, however there is at least no psychological reason why I am hearing this effect, and even less so for my wife who knows nothing about electronics and has no basis for not being happy with the sound other than what she is hearing.

I've constructed the '96 preamp very carefully with no modification other than to use bipolar electros in place of polarised ones in coupling positions (which while perhaps may not improve the measured performance, I hope you would agree should not decrease it significantly in this design?).

The tone controls are just brilliant, exactly what I was looking for, so I want to offer my profound thanks to you for providing a great design and my assurance that I absolutely believe the design is exemplary. However I am left with a problem because my sound system just isn't as enjoyable to listen to through the preamp. I really feel it is somehow compressing the signal so that the sound is somewhat constrained. I don't know why this is, and I think if you were to accept that it exists, trying to measure this effect would be extremely difficult.

I'm not at all convinced about the vital importance of the quality of DC power supplies beyond the usual basic precautions, the apparent black art of decoupling opamps or the evil of ground planes as some are suggesting, nor do I believe that the opamps themselves are to blame (for this anyway). But I am convinced there is something about the design causing this subjective effect.

I am directly comparing it subjectively to the "headphone amp" pre I posted earlier and there are many similarities bewteen the two (ie they copied much of your design ), the differences being:

1) there is no tone stage (which is why I decided to build yours)
2) the AGC feeds a discrete opamp buffer with inverting unity gain, which is very simlilar to your "blameless" power amp circuits but with a simple EF output stage with no drivers running heavily in class A, and two pole compensation.

This suggests two possible culprits for the "compression" effect:

1) The tone stage itself - seems unlikely since the effect is there with the tone cancel switch on also.
2) The AGC does not make a good output driver and a buffer after it really helps

While your measurements suggest at least that previous speculation about pot loading is not the cause of the latter, could there be something else? Perhaps the way feedback is applied around both opamps or something?

[owdeo]

Originally Posted by redjr

I thot of paralleling too. The only place in the article they mention the 10-pin ribbon cable refers to a connector on both ends. No talk of 2 connectors on one end, but it does make sense. I suspect that part will have to be homemade, as it might be very hard to source something like that. I managed to get an 11" piece with one connector on each end from eBay. Now all I need to do is crimp another connector on one end.

You can buy IDC crimping tools, but if you have a vyce in the shed that works just as well if you're not too heavy handed - just place the socket in the vyce with enough pressure to hold it there, insert the cable and line it up exactly parallel with the pins then squeeze it home. The important thing is to apply even pressure over the whole width of the connector when it crimps the wires.

[redjr]

Originally Posted by owdeo

You can buy IDC crimping tools, but if you have a vyce in the shed that works just as well if you're not too heavy handed - just place the socket in the vyce with enough pressure to hold it there, insert the cable and line it up exactly parallel with the pins then squeeze it home. The important thing is to apply even pressure over the whole width of the connector when it crimps the wires.

I ordered the female connectors today, and almost bought a crimping tool. I figured I could use my vice for this one off connection. Years ago when I was home-brewing computers I made my own IDE and parallel printer cables. I borrowed a handy crimping tool this guy made in a machine shop. It worked really well. But that was then. I can't wait to get this pre-amp operational - to hear how it sounds. Hope I soldered everthing correctly. It took a while to amass all the components, and the final two rotary switches have been shipped! Now if Elektor will only offer a spiffy enclosure sometime this year. Since I've already spent more than I originally planned for on this project, it begs to have a nice case and I'll happily pay! Well, maybe not happily.

[abraxalito]

Originally Posted by owdeo

I'm not at all convinced about the vital importance of the quality of DC power supplies beyond the usual basic precautions, the apparent black art of decoupling opamps or the evil of ground planes as some are suggesting, nor do I believe that the opamps themselves are to blame (for this anyway).

As far as I can see, I'm the only poster who has mentioned grounding. I did not say (nor do I believe) though that ground planes are evil so perhaps a little clarification is in order. A ground fill differs from a ground plane in that the latter is unbroken - a ground plane might have the advantage providing low inductance power, but only when used in pairs. A ground fill, used singly has no such redeeming quality.

If there's anything evil about ground fills its that they suffer from common ground impedance couplling and I've found 5532 opamps particularly susceptible to the induced noise on the signal ground. The only solution I know of to the effect of common ground impedance coupling is to use star earthing for signal ground and keep this separate from power ground.

Grounding Common Impedance Coupling (pdf download)

[sandyK]

You can buy IDC crimping tools, but if you have a vyce in the shed that works just as well if you're not too heavy handed - just place the socket in the vyce with enough pressure to hold it there, insert the cable and line it up exactly parallel with the pins then squeeze it home. The important thing is to apply even pressure over the whole width of the connector when it crimps the wires.

Don't forget that these home brew IDC cables can also cause all kinds of funny things to happen with DACs too.

Regards
Alex

[owdeo]

For what it's worth, here are my impressions of a couple more opamp types in the AGC section. Of course I am exaggerating the difference to some extent (no doubt some would say completely ) to put my impressions into words.

LM4562: (appears stable, no sign of oscillation on my 'scope). Cleaner treble, slightly less "air", flabbier bass that almost booms. Also a slight hard edge to the sound - not in a painful way, just one that sounds very "hifi" in an artificial way but detracts from the realism or believability of the instrumental timbre. The two speakers are more noticeable as the soundstage just doesn't seem to project out from the speakers much. I could not enjoy listening to any tracks I played with these in situ. Not for me thanks. Probably really good if you mainly like listening to demo CDs.

JRC 5532: Again completely different to the other two 5532 types I've tried now (surprisingly the degree of difference with the LM4562 was not any greater). These are very musical - there is something euphonic about the midrange that makes instruments sound natural and a pleasurable to listen to. Bass slightly fuller also. The sound also seems to project out of the speakers more somehow but not unnaturally forward. These aspects seem to come at the cost of the top end though, it seems a little coarser and not as refined. But they are so musical I just kept on enjoying every track I listened to and didn't notice the "sound" much.

The basic character of the preamp remained with each opamp change - I still feel it is somewhat "compressed", but it seems clear now to my ears that something as mundane the opamp brand can make all the difference to whether the preamp gets in your way of enjoying the music. I think I'll stick with the JRCs for now, the thing just makes music with these and that's the whole point after all. The On Semi's are close and perhaps slightly more "refined", but just not as enjoyable. But these are the only two opamps that don't seem to feck up the sound in this design so far.

[Mooly]

What you are finding out is that technical perfection and what you like to listen too are totally different concepts. You have to find what works for you.

I can't remember whether you tried a simple pot in place of the active stage (AGC had me confused for a second as it means something totally different ) but you should. As I mentioned earlier the tone controls in isolation are absolutely superb.

Without wanting to add too much the subjective/objective discussion all I can say is that certain circuits/opamps/topologies can sound far more pleasing to listen to than others. Absolute technical performance or excellence has no real bearing on this either. Some designs just make music, others do not. You have to see what works for you.

Just some off the wall ideas...

have you tried the OPA2604 as that can sound very "analogue" and musical.

also try biasing those opamps in the gain stage into Class A

[owdeo]

At the risk of getting into phillosophy I'm not sure that I entirely agree.

Yes it absolutely comes down to a matter of personal taste, but I think there's more to it than that. Being true to the original sound is what "hifi" is supposed to be about. Obviously if you only listen to music of electronic origin or artificially recorded music (ie pop/rock) the concept has no meaning and you can merrily change your components or whatever to achieve the sound you like.

But I'm looking for sound quality that gets me as close as possible to the real sound of an acoustic group, whether it be a small jazz ensemble or string quartet or a full symphony orchestra. This shouldn't be as much a matter of taste if you regularly attend concerts and play instruments.

I would say more that what I'm finding out is that the merit of the conventional set of measurements, eg THD curves and SNR etc could be missing the point somewhat in terms of how they relate to our perception of reproduced sound. These parameters are of course essential for verifying that a design is working properly and I am by no means suggesting they are irrelevant, just that perhaps there are other things we could or should also measure. This would make the whole thing awfully complicated though and it's easy to see why a designer would focus only these parameters.

I agree that I should try bypassing the Active Gain Control somehow - probably by bypassing the pot and buffer and keeping a fixed inverting gain stage at the output to keep the phase right and provide some gain. I'll need to put a pot on the input before the input buffer to control the volume then. I guess a 10k log should do.

I liked the OPA2604 in another conventional preamp circuit I tried it in, but not as much as the 5532, and also as an IV converter in a DAC output, where I think it was the best of all those I tried. Whether it's a good choice in this design seem doubtful but I guess I can try it and see.

[jan.didden]

Originally Posted by owdeo

[snip]I agree that I should try bypassing the Active Gain Control somehow - probably by bypassing the pot and buffer and keeping a fixed inverting gain stage at the output to keep the phase right and provide some gain. [snip].

What you could do is put in a toggle switch on the back of you preamp, invisible from the front, one position the AGC in a fixed position and the other pos a fixed gain or attenuation.
Then find someone who would randomly set the switch without you knowing it, and without he/she knowing which is which.
Leave the room, have the switch changed (or not), then the switchr leaves the room, you come in, and decide whther you think you hear the AGC or the fixed setting and note it down.
Make sure at no time do you have contact, even eye contact, with the switcher.
Do 100 trials and post the results.
That way you will have a strong point that there is sometinmg going on (or not), and people would be more willing to investigate it.

jan didden