Differential power amp - match components (Alesis RA300)?

[living sounds]

I had a channel of that amp die on me and several components got burned (roughly 'sprayed' in red in the schematics). Went on to replace all of them and it works again.

I've done measurements as good as I could:

First using big 4 ohm resistors the channels appear to be identical across the audible frequency range in amplitude up to high amplitudes measuring with scope and multimeter. No DC present.
I also measured THD at low amplitudes connecting the outputs directly to my soundcard inputs and set the bias on the newly repaired channel for lowest THD at these levels.

However, THD on the "new" channel was not quite as low as on the old one. Small deviations here are probably to be expected, but it also seems to me from listening that the "old" channel has somewhat better transients, high end response whatever, even though amplitude under load measured equal in level with sinewaves.

Now, I used 10% resistors for the damaged ones instead of the required (hard to obtain) 5%, and couldn't match them since measuring 0.22 ohm properly is not what my multimeter was designed for. I also didn't match transistors.

So, I wonder if - judging from the schematics - doing either would improve anything  here.

Thanks!

 

[JohnRoberts]

Have you re-tweaked the class A bias (VR4)?

You can adjust it for the same voltage drop across the .22 ohm emitter degeneration resistors, as in the good channel (with no audio signal playing, and no load if the amp has DC offset).

If you want to get objective You can look for crossover distortion at HF with a scope. Using a 10 khz sine wave, look at the zero crossing. Since bipolar transistors do not turn on or off instantaneously, designers introduce a small class A current to keep them both turned on just a little (tens of mA) down around 0V.

Too much class A current will waste a lot of power, and generate a lot of heat, so don't be tempted to dial in too much. With a scope you should see a significant visible improvement between the output stage starved (no class A) and 10-20mA of class A current.

If you can't look at it that closely, just adjust them to be the same as the other channel. (Hopefully you prefer the sound of the one that is biased correctly  8) ).


JR

 

[living sounds]

Thanks John!

Yes, I've tweaked the bias for lowest THD via VR4. I had the output connected to my soundcard and used a 1k sinewave at a low amplitude (so as not to damage my converter) for this (with an analyzer plugin).

So is there no reason to match the four 0.22R resistors or any of the power transistors to get better performance? I wonder why they specified 5% here.

I'll look into the crossover distortion, but my scope is pretty basic. Is a lot of amplitude required or could I use the soundcard for this (it will accept +/- 5V AC before clipping)?

 

[JohnRoberts]

Crossover distortion is an output device speed phenomenon so not as apparent at 1 khz as 10khz, and percentage wise somewhat hidden by high amplitude measurements, since the perturbation is only for several mV around zero crossing. Of course using a sound card, the higher harmonic crossover distortion caused by a 10 khz sine wave may be well above the sound card's response. 

This is one of the classic problems with blindly reading spec sheets. If distortion is specified at full output even with crossover distortion present it can measure as a very small percentage of THD, while it will be clearly audible while listening to low level music.

If you can't see or measure this, adjust it for the same idle current as the channel that sounds OK to you, but generally not much more than 20 mA or so.

While I have never tried this, you might be able to get around the sound card bandwidth limitation by using a two-tone IMD test. Using two closely spaced but in-band sine waves (I used 19 & 20kHz for my old DIY analog bench unit). The IM distortion product will be the difference or 1 KHz which should be captured by your sound card. 


JR

 

[living sounds]

Thanks! My converter (Lynx Aurora) goes up to 192khz samplerate, so no problem there. I'll measure at 10khz later today.

The thing is that even at the lowest setting the other channel still had less THD at 1khz.

And about those resistors - how would a mismatch affect the signal?

 

[JohnRoberts]

While I am not sure exactly which exact resistors, but most of the circuitry in that area is forcing current sharing between the power devices, establishing overload current limiting, etc.

There is overall NF that should be keeping distortion low (within reason).

None of the resistor values seem critical to me, as there is a trimpot to set actual class A current. If you replace power devices I would be more suspicious of that. While the current limiting only senses the current on one device, if really poorly matched (Vbe) they may not share equally at very low current.  I don't know what that basic amp is capable of. As I recall the Alesis was budget gear. Unless you are using low distortion load resistors (I'm not making this up) you may be seeing distortion related to the load resistors.

The Class A bias will improve crossover distortion, but there are limits to how much it can improve the total distortion, so in general stop increasing it when you see the major HF distortion drop.

JR

 

[ricardo]

Living, what is the bias current on the good channel?  ie what voltage do you measure across each 0R22 resistor.  If they are all the same, it shows the output devices are pretty well matched for Vbe.

And what is it on the repaired channel?

 

[living sounds]

Thanks everyone! With some problems on the road (pot trimmer breaking, resistor touching chassis and burning up ground plane etc. - note to self: don't do electronics when you're tired) I managed to get the amp working properly. I noticed the relay I had replaced got a little warm in use, so I resoldered the connections and things got better. Then I reset the bias on both channels and distortion is now below the noise. That's significantly better than stock, too.

The amp is sort of a mixed bag, it's a budget unit, but the mechanical construction and the overall design is pretty good IMO, and quite easy to work on (once it's disassembled), all through-hole. Nice and well shielded power transformer, too. The relay is garbage however, and the PSU pcb is of low quality and prone to breaking. Also, the schematics call for COG ceramics in many places but what actually got used are cheap disks. I changed a lot of caps for better ones where it makes sense, also used polarized electrolytics in the signal path, and this is now a very nice sounding amp.

In these latter revisions they had even substituted the often-failing TA7317P protection IC with a little discrete surface mount board. But it still got discontinued because of frequent failures. I guess at the price point they couldn't economically built something better. Now there's only Behringer, and their comparable amp has a severe design fault.