'AK Mods' SC Class A 15W Amplifier

[jeffc]

Well I've tested 2 different 4 ohm-specified full-range speaker drivers pulled from cars and no cone movement detected visually or with a finger resting on the cone... However, these single drivers will probably present a more benign load compared to speakers with an elaborate crossover... Still, no cone oscillation is consistent with experiences with the DML panels. So Sean.. I can't help pinpoint whatever is causing the oscillations, sorry.... my amp seems OK .... except for distortion when the volume is pushed too high...

Seeing you like your music a little more "feral" sounding, you could always sell your Class A to Phil at cost price !

G'day Alex,

Yep "feral" is me. However, Phil won't be able to afford cost price + the cost of the blood, sweat are tears I put into building this amp

Up front, it is a very sophisticated sounding amp, neutral and natural tone, lovely female and male voices, plenty of detail and many other terrific qualities....

Problem is I'm still not using a preamp, just a passive ladder attenuator, and the Class D amp with 47K input impedance, a THAT1200 line drive stage essentially acting as a built-in preamp, output filter tailored specifically for a 4 ohm load, oodles of power (250W into 4 ohm), plus tweaks I've just made to it that I'm still in the honeymoon period with, makes it the better option for the time being. For my 6 ohm Visaton B200 full-rangers in winged OB panels that I've re-put-back-together as rears for panambiophonic when I get there (......which has been my grand plan as you know for a couple of years now and still is... and as I'm increasingly struggling with stereo now listening solely via ambiophonic at my computer desk... ) the Class A should work fine.

It is an amp that requires very careful attention to detail, with the right upstream/downstream ancillaries, to get it to sing perfectly, and I as yet haven't got these. Have plans to though.

cheers.. jeffc

[Deleted]

my amp seems OK .... except for distortion when the volume is pushed too high...

Bloody hell Jeff, you must be even deafer than I am, or those panels of yours are VERY inefficient.
Have you had your ears syringed recently ?
My DCM QED 1A are only 87dB/1W/1M, but there is no way I can use anywhere near full power in a typical sized room without being deafened.

Kind Regards
Alex

P.S.
Thanks for doing those tests for Sean !

[Deleted]

Hi Jeff

Glad to read you again IMHO Alex's Class A really does benefit from being driven by6 an active Pre and if you can find a way of hooking up the Class A HA/PRE that you have then I’m sure that you will be much happier with the sound.

I think that I’ve mentioned that all of the passives that I’ve tried have sounded a little grey compared to Alex's HA/PRE and even the Aikido.

Could be worth experimenting with.

Hi Sean

Good work but now I think that it may be time to do some voltage measuring around the board.

I used MPSA18 for my first set up build and later went on to dual matched single units later.

So I had 26mV on my OP with speakers connected.

I thought that it was my OP correctors causing the offset problem but nope. After Alex's input we found that I had the MPSA in back to front which was maxing out the offset corrector IC. (I'm new to transistors so my fault)

But with the MPSA located the right way around the offset dropped like a stone. Jeff is right when he said that with the loudspeakers connected that the offset should drop still further so it could be worth hooking in the stunt speakers just to be sure that is happening.

Then I’d go for a voltage measure.

I have my offset corrector connected to the speaker side of the OP resistor BTW but it's hard to see how that could make much difference.

Could be worth a try though.

Take care

[sean]

With FB cap bypassed and no speakers, and 1k on input, DC output at speaker terminals is 2.7mV. Pin 6 at is at 376mV DC.

As above but with 8 ohm test speaker connected, DC output on speaker is 1.0mV. Pin 6 is now 396mV.

With a 4.7 Ohm 50W resistor connected, DC output on resistor is 0.9mV, Pin 6 is 394mV.

Looks like there's too much gain - the 4K7 and 1M resistors are definitely right though. I swapped op-amps for the other OPA134 and got the same results.

[Deleted]

Sean
I will double check mine again in the morning if possible, but I am pretty sure that I had -0.2mV in, and -60mV out.
As you have seen, Jeff has no problems, as do Shaun and Alan, although they aren't using Will's PCB.
As regards the connection being on the wrong side of the Zobel, perhaps due to PCB layout difficulties,
the Zobel is likely to have only a couple of 10s of milliohms resistance, so it shouldn't matter.
Do you have a TL071 ,AD744 etc. that you could try instead ?
Regards
Alex
P.S.
Do you have a series 220 ohm resistor at the inputs, instead of the original 1.8K specified.

[sean]

Hi Alex, thanks for that. The only other single channel op amps I have spare are 741 and 3140 op-amps, are either of those any good to try?

EDIT - just found one AD797 too, which is best to try do you think?

[Deleted]

Hi Sean
They aren't suitable.
Goodnight from Sydney Au

[jeffc]

Alex,

Distortion was with the crappy 4 ohm car speaker drivers I tested, the Class A plays my DML panels plenty loud without distortion, enough to get me into trouble with those wanting to watch TV upstairs.

Hi Shaun, from all that's been discussed, I'm sure a pre will help heaps. I might have to reorganise my Class A HA for this....... when I get some time, retirement/redundancy not too far away

Main problems with the Class A in my Grotto setup as it exists now minus preamp are the HF, which while sounding as "right" as I've heard, is far too recessed to energise the music, with the soundstage collapsing because of this to exist in a hole between the panels, and there is also something a little strange about channel separation that diminishes the sweet spot area, with relatively minor shifts side to side gluing sound to the panel, which is weird because the panels connected to my Class D Audio amp have such a broad and uniform off-axis response and wide soundstage. These problems could be partly be my own making with capacitor and other component types selected for my build, no preamp, the Class A wondering WTF its been connected to (naked piezo elements for example ) and probably bits of all that will take quite a bit of fiddling about to get right. Plus I still have some transformer buzzing in the PSU that while not intrusive with music playing, I'd like to fix.

cheers.. jeffc

[Deleted]

Hi Jeff

On using the HA as a pre

You could make a quick and dirty ¼ inch /two RCA Y splitter just to have a listen. On channel separation well that’s a strange one. As I remember you have a dual mono rig with the only common connection being the ground.

So that leads me to the question

How is your amp grounded and have you tried a different front end just to rule that out?

''capacitor and other component types selected for my build''

That could be worth comparing notes on. Especially what caps you have used where. I’m having a bit of a quiet fiddle with the caps in the mule which has been fun .

Also what value inductor are you using in the OP zobel?

Transformer buzzing

Mechanical?

On your panels

Speaker positioning may also help.

that's an area that my PMC are a bit picky about to say the least.

I’m getting a pretty 3d soundstage TBH with the PMC FB1+ but they are really easy to drive/go loud and present quite a flat 8ohm load to the amp.

Just as a point of reference what cables do you have in ATM?

Yup all grandmother and eggs conversation but I’d be interested in having it all the same.

Take care

[Deleted]

You could make a quick and dirty ¼ inch /two RCA Y splitter just to have a listen.

Hi Shaun and Jeff
I actually have one of those made up, using cheap dual RCA sockets from Jaycar and a 6.35mm Headphone plug that Mike sent me a few years ago,
and we used the Class A HA this way at a listening session some time back.
A couple of us have found that the HA when used as a preamp actually sounds best with 100 ohm series output resistors.

Kind Regards
Alex

[Deleted]

''A couple of us have found that the HA when used as a preamp actually sounds best with 100 ohm series output resistors. ''


Hi Alex

Yes I’ve found that 100R sounds best when using your PRE and also a little more controversially I’ve found that 100R sounds pretty good for the HA into my K701 headphones.

But that's just to my ears in my set up IMHO.

ATM I have your HA connected to my Opus/IVY 3 via 1/4 inch jacks (a very short length of VDH 102 MK2 cable used) which works quite well. I have the 1/4 inch socket wired up to disconnect the RCA IP and OP when the 1/4 plugs are inserted which allows for multiple use and quick comparison between components. I have the Opus/IVY3 set up the same way.

Looks a bit telephone exchange like but I’d love to get my hands on some of those GPO plugs and sockets which where top draw.

Anyhow

I’ve been able to run the Mule and my existing HA/PRE simultaneously that way which has been more than useful for comparing changes.

Take care

[Deleted]

Hi Sean
It would appear that the figures I quoted with the posted diagram for the Offset Corrector were for the HA/Preamp taken some time ago..
The exact same Offset Corrector design is also used with the 15W Class A, and others haven't had any problems with either the preamp or 15W Class A .
I just took a series of readings on my 15W Class A with a 1K resistor across the input of the L channel.
With a 3.6 ohm load. (2 x 1.8 ohms in series) , the DC Out was 1.5mV which is pretty good for a Power Amplifier .
Pin 6 of the OPA134 was 190.6mV . With an 8.2 ohm PW5 load , the voltage at pin 6 was 193mV ,and the same with no load.
The output of the Offset Corrector was around -6.4mV. The readings were taken with the cover off, and the amplifier hadn't been warmed up,
due to chopping and changing resistors etc. All readings were steady.
We may need to get more voltage readings from you to try and solve this. What is the DC resistance of your speakers ?
Perhaps there is something in your speakers that makes them a difficult load for many amplifiers as reported on the Internet.
Have you double checked that the output of the correctors are via 10K and 75K series resistors ? 75K is not a common value,
and a few have had resistors supplied over the counter that were incorrect.
Kind Regards
Alex

[Deleted]

with the soundstage collapsing because of this to exist in a hole between the panels, and there is also something a little strange about channel separation that diminishes the sweet spot area, with relatively minor shifts side to side gluing sound to the panel, which is weird because the panels connected to my Class D Audio amp have such a broad and uniform off-axis response and wide soundstage.

Hi Jeff

Are the photos that you posted here rockgrotto.proboards.com/thread/7461/ak-mods-class-15w-amplifier?page=43 the latest photos ?
In the last photo you don't appear to have connected the incoming PSU earths of both channels together. IIRC, you did connect the earths of the individual channels to their own heatsink though
because of minor hum problems.
It's also a good idea to connect these commoned earths to a banana socket on the rear panel , so that you can connects a lead with banana plugs to the same scenario on the preamp's case .
As Shaun will verify, this also results in a further improvement in imaging and SQ.
In the original S.C. article you will see the incoming 0 volts ("earth") lead connected to the baseplate, then to each channel.

This incoming earth is also connected to I.E.C. mains earth at both 0 volts lines of the PSUs in the PSU case . In this case, we should have both PSU PCB's 0 volt lines also connected to I.E.C. mains earth. .
I realise that we are using separate PSUs for each channel, but I believe that these earth leads should still be connected together for best results.
I have absolutely unbelievable soundstage imaging with mine, with the soundstage coming right up to the listening position on good material..
If you pass through Sydney again en route to or from Geelong, feel free to drop in.


Kind regards
Alex

[Deleted]

Hi Alex/Jeff

I agree with Alex that the grounding scheme is important which is why I mentioned it in my previous post.

I have my grounding set up pretty much as Alex has outlined and always fit the extra binding post on the back panel as standard as it allows a little room to experiment without too much fuss.

Connecting the ground lines together will ensure that each channel has an identical potential to ground which may solve your imaging problems.

As an experiment you could try measuring the mV defence between the OV lines on each channel (just put a MM across the ground lines)just for the sake of interest.

For the little time and cost involved(a binding post and a bolt) a slight rearrangement of the grounding scheme could be worth a go.

Take care

[sean]

Hi Alex,

DC resistance of my speakers is 3.3 Ohms. DC resistance of the test speakers is 6.4 Ohms FYI. And if it's relevant at all, i'm using 3m of Naim speaker cable, which has LCR characteristics of 5uH/50pF/75mOhms for the 3m length.

Could it be that the 3n3 cap in parallel with the 1M resistor is causing problems? What if I increased the value? The 10k and 75k resistors are both OK, I measured them with a multimeter as well as checking colur codes (the 75k measures 70k when left in-circuit but it's close enough isn't it? I guess something else is affecting the reading a bit). What's the 100n cap to ground between them doing exactly? Is that likely to be causing some oscillations?

[Deleted]

Stuffing about to begin....tadaaaaah!

What's the current wisdom on either, fitting the OPA134s directly to the board or use a DIP plug?

[Deleted]

Apr 19, 2013 15:05:54 GMT @cjarchez said:

Stuffing about to begin....tadaaaaah!

What's the current wisdom on either, fitting the OPA134s directly to the board or use a DIP plug?

Hi Chris

Good question which will draw out different opinions I’m sure.

Myself I always use sockets with my IC’s partly due to lack of confidence in my own work and partly due to them being a total PITB to get out of the board should things go wrong. Also it’s useful when firing up the amp for the first time to have the IC removed and the FB cap in place to allow checking of the 15V regs (er I think).

Others have pointed out potential problems when using sockets but TBH I’ve not come across any of them.

Other opinions please.

Good luck with the build Chris

I’ll be routing for you.

Take care

[Deleted]

Hi Shaun,

You echo my own thoughts and experiences. I've always had the intention of, once everything checks out remove the sockets and pop the IC's straight in. Which, of course, I've never done!

Thanks for the cheer leading

[Deleted]

''Which, of course, I've never done!''

GOOD MAN


I was having a look around the web for some info on your Sybils and it seems that the specifications that you have posted for them are for the passive version.

Good news as not having to drive that cross over and going active to the driver should give you a little more sensitivity not to mention punch and drive.

Great looking speakers and from memory great sounding as well.

Take care

[Will]

Sorry to hear that you have some problems Sean. The advice you are getting at the moment is spot on, and can't offer any more myself.

You are right that the input for the offset correct should come from the output, and it did i the HA/pre version of the pcb, which this PCB is based upon. Looks like an oversight my part when including the zobel on this PCB. Still, as Alex says, it should make little or no difference in operation.

Hope you crack it!

[Deleted]

Sean
What has me beat here is that you have around double the voltage out from your correctors that I measured with mine.
That suggests that there is more correcting needed . We may need to do voltage readings around the front end and VAS stage.
The offset corrector will work without either the 22nF or the 100nF, which reduce wideband noise generated by the opamp itself due to the very high gain used,
Gain can be reduced by changing the 1M to 470K, but I doubt that this is the problem.
Regards
Alex

[Deleted]

Apr 19, 2013 15:32:44 GMT @cjarchez said:

Hi Shaun,

You echo my own thoughts and experiences. I've always had the intention of, once everything checks out remove the sockets and pop the IC's straight in. Which, of course, I've never done!

Thanks for the cheer leading

Hi Chris
I would recommend that with multi layer PCBs that I.C. sockets be used for areas such as this. The magazine type PCBs usually provide greater areas of copper around components,
realising that not everybody is experienced. It's an entirely different matter when the process is automated, and errors do not normally occur.
Kind Regards
Alex

[jonclancy]

DIP Socket or no DIP Socket?

I've used a socket. Primarily for conveneience in swapping out duff OPA and for testing (And by coincidence I had a duff OPA - for the first time ever!!!).

Also, I might try the 100R output resistor. Although I'm currently using the HA as an HA, I'll be using it as a pre as well when I get my Class A amp built up. Nice tip - thanks Shaun.

[Deleted]

Hi Jon
Any chance to follow up on the nudey resistors yet ?
BTW, with that little delay PCB that I am using in the preamp, it should be possible to do without a toggle switch to change between headphones and RCA sockets
by taking the earth from an amp PCB's output, and connecting it via the sleeve break contacts of the headphone socket, to the delay PCB as it's earth feed.
In other words, plugging the headphones in will stop the delay relays from operating. That may be easier than breaking the A.C. supply to the Delay PCB.
If you can tolerate a slower turn off of the output to the PA, -VE DC could be used instead of A.C.
Another option could be a front panel toggle switch to switch between headphones and RCA outs, doing away with the delay PCB and relying on the Speaker Protection's delay .
I would be inclined though to use a 4pole double throw toggle switch with parallel contacts, or even a 2 position 6 pole rotary switch with parallel contacts, as toggle switch contacts can go high resistance eventually. Spare contacts on a rotary switch could be used to terminate the earth wiring of the output leads..

Regards
Alex

[Deleted]

Apr 18, 2013 12:47:06 GMT sean said:

With FB cap bypassed and no speakers, and 1k on input, DC output at speaker terminals is 2.7mV. Pin 6 at is at 376mV DC.

As above but with 8 ohm test speaker connected, DC output on speaker is 1.0mV. Pin 6 is now 396mV.

With a 4.7 Ohm 50W resistor connected, DC output on resistor is 0.9mV, Pin 6 is 394mV.

Looks like there's too much gain - the 4K7 and 1M resistors are definitely right though. I swapped op-amps for the other OPA134 and got the same results.

Hi Sean
Thinking about this a little more, as your offset corrector needs to put out about double the voltage mine does, and yet you can't get down to a fraction of a mV, suggests that the offset corrector is working hard to correct excessive DC offset. I think we may need some front end readings with the voltages across the LTP and CM emitter resistors, as well as the collector voltages of the CM transistors, and the top and bottom resistors of the VAS stage. The attached should give a general idea, although the readings will be slightly different due to the 200 ohms instead of 220 ohms.
Regards
Alex