suggestions for improvement

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Just another Moderator
Joined 2003
Paid Member
Hi guys,

I'm attaching a scan of the circuit diagram of my power amp. Note this is actually an integrated amp, but I'm considering ditching the preamp stage and just using the power amps. Any suggestions on how to improve on the design (it was a kit amp, I didn't design it) would be appreciated.

Apparently the pcb layout is critical, with small variations greatly affecting distortion performance (quoted performance is about .01% THD), so I guess I'm limited to replacing existing parts, with maybe a few options to add additional components (or remove) on the reverse side of the pcb if warranted.

Also input of 1.5V is supposed to produce max power.

The power supply rails are actually closer to +-63V, its a 300 VA 45-0-45 torroidal, with 16,000 uF (2 X 8000 uF) on each rail.

Also I think that all of the 1n4148 diodes are actually 1n914's

BTW I know nothing about amp design :goodbad:

regards,

Tony.
 

Attachments

  • amp.jpg
    amp.jpg
    58.1 KB · Views: 1,420
The first thing I'd think about is current mirrors in the input stage. That will increase open loop gain and improve differential balance. You'll have to play with the compensation a bit to stabilize it... All in all, it's already a pretty straightforward and functional design. You might want to think more specifically about what it is you're trying to improve.
 
Just another Moderator
Joined 2003
Paid Member
Thanks SY. I guess what I'm after is if anyone sees anything nasty with the design then point it out. The main problem I have had with the amp, is noise, which I think may be in the preamp stage.

I'm going to go out and get a soundcard preamp kit today, which with some software allows you to use your pc as a CRO, and a book on using oscilloscopes. Maybe after playing around a bit I'll have a better idea what I want to achieve :)

Regards,

Tony.
 
That looks like the textbook circuit from the Hitachi application notes for the J49/K134 parts.

Replace any ceramic caps with film types.

I would replace the 47µF electrolytic in the feedback loop with a high quality non-polar type as high a voltage that will fit on the board and tack a 0.1µF film type in parallel.

The 0.15µF on the speaker output should be a very high quality type, I would use a stacked film and foil type of at least 250V here. If this cap goes bad the amplifier can sound 'noisey' due to oscillation.

This amplifier is old enough to justify replacing the 100µF and 22µf power supply bypass caps with new ones.

You could also try replacing the main power supply bridge rectifier with some Schottky diodes.
 
Just another Moderator
Joined 2003
Paid Member
djk said:
That looks like the textbook circuit from the Hitachi application notes for the J49/K134 parts.

Replace any ceramic caps with film types.

I would replace the 47µF electrolytic in the feedback loop with a high quality non-polar type as high a voltage that will fit on the board and tack a 0.1µF film type in parallel.

The 0.15µF on the speaker output should be a very high quality type, I would use a stacked film and foil type of at least 250V here. If this cap goes bad the amplifier can sound 'noisey' due to oscillation.

This amplifier is old enough to justify replacing the 100µF and 22µf power supply bypass caps with new ones.

You could also try replacing the main power supply bridge rectifier with some Schottky diodes.

Thanks djk,

Yep it is old in fact about 16 years old! And you are right it is based on the Hitachi application notes. I was suspicious about the .15uf in the RLC network, after reading this thread http://www.diyaudio.com/forums/showthread.php?s=&threadid=19327 The original spec was for a phillips MKT-P 2222 330 40154, but I think from memory they substitued something else (another type of mains suppression cap).

I'd allways wondered about the rectifier bridge It's one of those four pin in a metal case type one's not made from discrete diodes.

What is the .1uF in parralel with the 44uF for?

regards,

Tony.
 
Based on some published work by D.Self the 47µF electrolytic in the feedback is a potential noise source (Distortion mechanism #8 of 10 from is list). His "fix" if you are willing to made mods is to increase the value to 220uF - 470uF. It's function is just a DC blocker to protect the foodback loop from any DC offset at ground. Apearently when you raise the value of an electrolytic sufficiently any distortion is confined frequencies beyond audio. This shouldn't have any other effect on the feedback loop.

If you can find a local audio tech with a distortion analyser, have him check and adjust the bias. Supossedly SS amps will drift with age.

At this age some electrolytic capacitors maye be getting a bit going south. If not yet they will. If it were me I would replace them. I don't think anything exotic is needed, just new.

I woould also add snubbing caps to the bridge rectifier (.1uF ceramic high AC voltage). This is easy, cheap (as in pennies) and does no harm and as I don't expect this amp design is all that great in the PSRR category it might even help.
 
Just another Moderator
Joined 2003
Paid Member
Thanks Sam,

found the reference: http://www.dself.dsl.pipex.com/ampins/dipa/dipa.htm

Interesting reading. (I've only read the bit about the negative feedback section so far, I supposed to be doing some work that I brought home for the weekend)......

One thing I'm pleased about is that my amps circuit board seems to fit with the star earth topology, allways wondered why it looked like that :) and that the 47uF cap in the feedback circuit is connected to its own branch from the central earth, and not sharing it with any thing else at all.

I didn't know what snubbing caps were so did a bit of a search and came up with this: http://mitglied.lycos.de/Promitheus/components_and_schematics.htm

Is the power supply in fig 3 Normal? I haven't ever seen two rectifiers used like that for +ve and -ve rails. Have seen separate transformers and rectifiers for each channel, but not dual rectifiers on a single +- supply. Would using two rectifiers like that drop the output voltage slightly? Also are the 10uF and .1 uF caps parraleled with the 10,000uF caps a good idea?

I think I'll take your advise on replacing all the electros too, while I've got the thing apart I might as well do it, saves the hassle later of finding which of them has blown :) when it does eventually happen.

Regards,

Tony.
 
Unfortunately you cannot replace the two diff pair collector loads with a current mirror, because the stage current effectively sets up the voltage reference for the differential voltage amplifier current.

In any event you would need a fully symmetrical current mirror.

This circuit is a good one and works very well. It was used for decades in high power amplifiers as it is very stable.

There is not much you can do to improve upon it, although you can replace the diff pair feed CCS with a simple resistor. This will improve sonics somewhat.

Cheers,

Hugh
 
Just another Moderator
Joined 2003
Paid Member
AKSA said:
Unfortunately you cannot replace the two diff pair collector loads with a current mirror, because the stage current effectively sets up the voltage reference for the differential voltage amplifier current.

In any event you would need a fully symmetrical current mirror.

This circuit is a good one and works very well. It was used for decades in high power amplifiers as it is very stable.

There is not much you can do to improve upon it, although you can replace the diff pair feed CCS with a simple resistor. This will improve sonics somewhat.

Cheers,

Hugh

Thanks Hugh,

Redesigns were pretty much out of the question anyway. It's nice to know that the basic design is good.

Eventually I'll probably end up using the amp to drive my two Subs that are in the planning stages, and make a new amp for the satelite speakers, but for now it still has to perform all round duties, driving my current 3 way speakers for both music and TV :)

I guess part of the reason for starting this thread was to get a feel for whether or not the amp had a sound design or not, and I think that from the feedback I've got, the answer is pretty much yes. (no puns intended :goodbad: )

Regards,

Tony.
 
I find the design rather strange, normally the high gain and extra phase shift from 2 series diff pair gain stages would require more complicated compensation. This amp heavily loads the first stage with the Rpi of Q4&5. Rpi = Hfe * Gm, both terms vary with operating current, although high voltage transistors can have fairly flat Hfe vs Ic. Few designers use circuits that so directly depend on Hfe linearity, multiple point matching of Q4,5 with a curve tracer would be necessary for acceptable performance.

The 15 pF “compensation cap” is a joke, the C seen at the input of the FETs is 400-1000+ pF, depending on Id (which determines the output stage gain and how much of the ~10 nF of combined Cgs, Cds is reduced by the follower gain bootstrapping) and this FET input C determines the dominant pole of the amplifier. All of the output FET capacitance is voltage dependant, causing a nonlinear dynamic gain when driven by Q4,5,6 Gm stage current output. Further, the high Cload on the gain stage requires large current swings in Q4,5 to move the output at higher frequencies; the current swing modulates both Gm and Hfe of Q4,5 and hence the load seen by the input stage, causing major gain modulation throughout the amplifier.

Since the problems are largely dependant on signal slew rate this amplifier may be quite adequate as a subwoofer driver but I would hesitate to call it a good general purpose audio amp design.

As for simple tweaks, I don’t see too much that wouldn’t count as redesigning the amp; for a sub amp I would think that good PSRR would be important, so I would keep the input ccs bias, I second (or 3rd?) increasing the 47 uF feedback, the Electronics World Capacitor Sound articles suggest that non-polar electrolytics have lower distortion.
 
Just another Moderator
Joined 2003
Paid Member
Thanks jcx.

Most of what you said is beyond my comprehension level, but I'll take your word on it :)

The amp definitely has some problems in the preamp design (It's actually an integrated amp, I only showed the power amp), so It doesn't surprise me that there may be other problems.

I guess what I really need to do is get off my behind and make my new speakers (which I'm sure are a weaker link at the moment than the amp), then do some A B comparisons with a known good amp, so I can see whether I can hear the difference, and then decide what to do with it.

Regards,

Tony.
 
This post is not necessarily a suggestion for you, Wintermute but more of a question for jcx, aksa and the other very knowledgable SS engineers here.

The 12k/15nF combo in the 2nd LTP doesn't appeal to me. What do you think about replacing it with another BF469, base connected to ground?

Other mods I might try: (1) increase tail current or tail resistors slightly in 1st LTP, in order to increase Iq in 2nd LTP, (2) increase the 100nF shunt cap across the bias pot (substantially).

I'm keen to hear the forum opinions on these proposals.

Besides these "ideas", I would personally replace all caps: larger psu caps with poly by-passes, much larger NFB cap (NP type); replace CCS with resistor. Most of these suggestions have already been made...
 
Just another Moderator
Joined 2003
Paid Member
DrG said:
replace CCS with resistor.

This one may not be such a good idea (I'm only going on what the original amp designer said here not on my own wisdom :) )

According to the writeup about the design, that came with the kit, the ccs is there to substantially improve the PSRR of the amp, so getting rid of it might improve sonics at the expense of hum????

I've bought a soundcard preamp (which I haven't built yet) and downloaded some CRO software, I'd like to do some measurments around the place (once I work out how :) ) and do some comparisons before and after changing caps etc as suggested.

Regards,

Tony.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.