'AK Mods' SC Class A 15W Amplifier

[Deleted]

Apr 23, 2013 19:36:27 GMT @cjarchez said:

Great news that Alan's amp is now driving 6ohm 'speakers well with an easy fix.

Could someone explain the "Cfb" to me. Why do you link it out or conversely pop in a 100uf cap?

Cheers

Having just got out of bed, I haven't got that far down the posts from overnight, but that's great news'.
When an amplifier is used with a feedback capacitor in circuit, the gain is reduced to unity at D.C. .
Without the capacitor in circuit the gain is still the same as set by the feedback resistor network, or 16 times here.
Kind Regards
Alex

[Deleted]

Apr 23, 2013 19:03:07 GMT sean said:

Hi Will, yes i've got some new opamps on the way, should have them tomorrow - in the meantime I need to test the C/B/E voltages on the 327 and 337, and see how they are. Does anyone have measurements from their working amps that I can compare my readings to? Cheers for all the ongoing help!

edit - the op-amps arriving tomorrow are TL071, are there any mods to the correcor that I need to make to use them or are they just drop-in replacements for the OPA134? They don't sound as good from what you guys say, but they're just cheap bits to get the thing working properly.

Hi Sean
You may have answered this already, but if the offset drops by around 16 times with the opamps still plugged in, and the FB capacitor fitted, then the offset corrector is working,
and you possibly have a fault caused by working into 4 ohms with the PCB error. The TL071 are compatible but noisier, with often less precise DC offset in the opamps themselves.
Is this the same with both amplifiers ?
Please, no deep and meaningful technical questions answers until I am properly awake,(even then ) and managed to get breakfast. Sometimes, due to overnight questions or emails that may be up to 2 hours after I get out of bed!

Kind Regards
Alex

[Deleted]

Apr 23, 2013 17:13:36 GMT drymdrum said:

Hi Alex/Shaun/et al,
did the mod on Will's boards, replaced the singed 10Rs and on fire up
had the pleasant surprise of music ....no good music coming out of the Wharfedale 6 ohms.
So that's a result, thanks Alex, not a difficult work around to do, I didn't lift the output block as I couldn't see
the point in making the job difficult.
Any point along the offending trace would be good to make the break which suited me, I suppose if you
had aspirations to submit your work for a gold star you might try to hide it under the block.........................
it's the music that counts for me.
Thanks again to all concerned in this project the music is top notch.
Regards,
Alan 
 

Hi Alan
I don't blame you for doing it that way. It wasn't that easy a job, even with a solder sucker and Gootwick, for me to safely remove my terminal blocks.
I would expect that you now have greater power reserves into the Wharfedales, and possibly sounding a little better too.
Kind Regards
Alex

[Deleted]

I have taken some additional measurements directly from my existing version.
Now that the PCB has been corrected, these measurements may assist with fault finding, although the component designations are different with this version.
I did not attempt to read the VBE of the BC327 and BC337 because of their heatsinks getting in the way, and the risk of damage. I would expect close to 600mV
The offset corrector is external in mine, but the schematic is the same. The voltage between the bases of the BC327 and BC337 (bias voltage) was 1.616V
This will change a little when the top cover is in place, as will other measurements, especially the bias voltage which will rise to close to 1A when warm.
The voltage at Pin 6 of the OPA134 was 189.3mV
Regards
Alex

[Deleted]

I discussed the issue of the take off point to the input of the offset corrector with a friend in Sydney who is a qualified E.E. Ionwyn ("owdeo" from DIYAudio) feels that the take off point in Will's PCB is fine.
He felt that it may result in a little less distortion from that point. However, considering that the offset corrector has 85K of resistance in it's output, and the correcting DC changes are fairly slow due to the 2 time constants at the input to the offset corrector, I do not believe that there will be any real difference between either take off point.
Regards
Alex

P.S.
With the Class A HA/Preamp, the input to the offset corrector is taken from the same point as Will has done here.
In the case of my present 15W Class A , it was more convenient to use the speaker terminal block for this connection, as the offset corrector is a separate small PCB.

[Deleted]

Hi Alex,
I'm just about to deliver the #2 build to my son it certainly sounds good down here so I'm sure he will be delighted with it.
As to take off point for the offset that's fine then to leave as is on both builds, we all appreciate your "mother hen"
concern for our Audio well-being.
Regards,
Alan

 

[Deleted]

Apr 24, 2013 11:02:33 GMT drymdrum said:

Hi Alex,
I'm just about to deliver the #2 build to my son it certainly sounds good down here so I'm sure he will be delighted with it.
As to take off point for the offset that's fine then to leave as is on both builds, we all appreciate your "mother hen"
concern for our Audio well-being.
Regards,
Alan

 

Hi Alan
I hope he realises all the hard work and huge numbers of hours you have put into it, let alone the component cost .
It's not exactly inexpensive to make.
Kind Regards
Alex

[Deleted]

Ah! what's a few shekels when your enjoying yourself,

Regards,

Alan 

[sean]

Here's more measurements from my output stage - 1k input resistor, 4R7 output load, wire link in CFB and op-amp fitted (DC offset at output still +77mV and pin 6 voltage is +13.9V btw - and the junction where the corrector feeds back to the amp is +9.0mV):

Voltages across 100R base resistors = 1.6mV, 1.7mV
Voltages across 180R collector resistors = 613mV, 621mV

Voltages across sets of 4 x 1R resistors = 248mV, 252mV (so bias current is 1A, more or less)

Voltages on BC337-40 C/B/E w.r.t. ground = 19.8V/0.81V/0.33V
Voltages on BC327-40 C/B/E w.r.t. ground = -19.8V/-0.67V/-0.17V

Also, the 302 output transistor is running about 10-12 degrees C warmer than the 281 if that's relevant.

[Crispy]

Apr 24, 2013 11:02:33 GMT drymdrum said:

Hi Alex,
I'm just about to deliver the #2 build to my son it certainly sounds good down here so I'm sure he will be delighted with it.

Well done Alan glad to see everything is sorted, I am pretty sure your son will be blown away with the sound

Apr 24, 2013 11:02:33 GMT drymdrum said:

we all appreciate your "mother hen" concern for our Audio well-being.
Regards,
Alan

 

X2, thanks Alex for all the help that you have given me up to now. (More will be needed in the not too distant future)

Regards

Chris

[Deleted]

Apr 24, 2013 13:46:04 GMT sean said:

Here's more measurements from my output stage - 1k input resistor, 4R7 output load, wire link in CFB and op-amp fitted (DC offset at output still +77mV and pin 6 voltage is +13.9V btw - and the junction where the corrector feeds back to the amp is +9.0mV):

Voltages across 100R base resistors = 1.6mV, 1.7mV
Voltages across 180R collector resistors = 613mV, 621mV

Voltages across sets of 4 x 1R resistors = 248mV, 252mV (so bias current is 1A, more or less)

Voltages on BC337-40 C/B/E w.r.t. ground = 19.8V/0.81V/0.33V
Voltages on BC327-40 C/B/E w.r.t. ground = -19.8V/-0.67V/-0.17V

Also, the 302 output transistor is running about 10-12 degrees C warmer than the 281 if that's relevant.

Sean
Are you able to put these measurements on a copy of the schematic and post it so that other members can check them too ?
Just scribbled readings would do. If both output devices are properly heat sinked they should have virtually identical temperatures.
Clearly, a fault condition has developed due to the PCB error and running into speakers with less than 4 ohms DC resistance.
After correcting the PCB error, Alan's amplifier had about 1mV DC out.
Do you get the same DC offset problem with the other amplifier ?

Voltages on BC337-40 C/B/E w.r.t. ground = 19.8V/0.81V/0.33V
Voltages on BC327-40 C/B/E w.r.t. ground = -19.8V/-0.67V/-0.17V

These figures are of concern. Are you able to measure directly between base and emitter of both ?
You will need to remove their heat sinks to safely do this.

Regards
Alex

P.S.
Given that one output device is running considerably warmer than the other, and the variance with BC327 and BC337 readings,
it suggests as Shaun had previously suggested, damage to the one or both of the BC327 and BC3378 driver transistors.
If there is no mother obvious damage, perhaps these 2 devices should be replaced to see if that rectifies the problem ?

[sean]

Hi Alex, those figures for the 327/337 were measured off of the terminals of the base side of the base resistors, and the relevant legs of the output transistors themselves, i.e. linked to the actual pins I wanted to measure by just a short PCB track, it saved faffing about with taking off heatsinks (they've got silicone paste on them). Both output devices are most definitely properly heatsinked, they're firmly bolted directly onto the 3U heatsinks of the Modushop case and liberally pasted with proper insulating kits.

So going off the figures above, B-E voltage on the 337 is 0.48V and the 327 is -0.50V.

I've only been testing one amp so far, but I did try the other amp on the 4 ohm speakers too because I thought the issue was only with one amp and that i'd done something wrong with the first amp originally, so that's probably problematic now as well. I'll link out the GNDLIFT and take the input to the corrector from the output side of the zobel, and then do some more checks. I figured that having one amp to fix was complicated enough without me posting even more data from a second identical amp for now... But yes, i'll see how it compares.

edit - I might get some more 327/337s in that case, it's easy enough to try and it'll save me doing more measurements. BTW, my op amps seem fine, I swapped in a TL071 today and it worked just the same as before.

[sean]

...been looking at CPC/Farnell but they only have the 327-25 and not the 327-40, would the 25 variant be OK to use? Specs are here:

cpc.farnell.com/on-semiconductor/bc327-25zl1g/transistor-pnp-to-92/dp/SC12768

The 337-40 is in stock though:

cpc.farnell.com/on-semiconductor/bc337-40/transistor-npn-to-92/dp/SC06664

I got the original 327-40 and 337-40 from Mouser, I can do again if need be but it's a bit of a pain having to make up an order to reach the free shipping amount of £50.

[Deleted]

Hi Sean
I did expect the OPA134 would be O.K. .The give away here is the hot O/P device and the very low VBE figures for the BC327 and BC337 which I would expect more in the range from 590 to perhaps 650mV if working correctly . I measured the tiny voltage drop across the 100 ohm base resistors to get some idea of their base current, as I didn't want to remove glued on fan heatsinks to measure VBE.
The voltage drop across the 180 ohms of the output devices is likely to be different too and may explain why one is much hotter than the other..
Yes, the 25 variants would be fine.
I am spoiled in that I can get this kind of stuff over the counter from the many Sydney Jaycar stores, and Altronics is only 10 minutes drive away, and Jaycar's HQ store
\is even closer..
Regards
Alex

[Deleted]

Apr 19, 2013 15:41:00 GMT @shaun said:

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.

I've always loved them Shaun, these are my second set after selling the first many moons ago. Why, why, whyyyyyyyy!

I asked at Naim what the impedance becomes when run actively;

All SBL drive units are 8 ohms, it’s the passive crossover that drops the impedance and 6 ohm is the minimum

was the reply.


That makes life easy then

[sean]

I replaced the output drive transistors, and so far it all seems OK now - output offset has come down to about 1mV (I can live with that!) under no load and about 0.1mV with a speaker load. Seems to play music well enough into the 4 Ohm speakers too. It's early days yet though, I need to get the second amp going now but that should be easy enough.

Now that the speaker ground is properly grounded directly, should I leave the ground lift resistor in place for signal ground or should I just link it out? It's linked out temporarily right now, and there are no problems at all with hum or anything. How was the 10R resistor causing problems originally out of interest, was it affecting the zobel network stability at all?

[Deleted]

Apr 25, 2013 14:29:59 GMT sean said:

I replaced the output drive transistors, and so far it all seems OK now - output offset has come down to about 1mV (I can live with that!) under no load and about 0.1mV with a speaker load. Seems to play music well enough into the 4 Ohm speakers too. It's early days yet though, I need to get the second amp going now but that should be easy enough.

Now that the speaker ground is properly grounded directly, should I leave the ground lift resistor in place for signal ground or should I just link it out? It's linked out temporarily right now, and there are no problems at all with hum or anything. How was the 10R resistor causing problems originally out of interest, was it affecting the zobel network stability at all?

Hi Sean

Phew

Glad to see that you've got the amp running with the expected low DC offset of less than 1MV under load.

I’d like to thank you for sticking with it when it would have been easy to give up and believe me I almost reached that point with my build,

Your work will help others further down the line which is what DIY is all about IMHO.

On the ground lift errrr well if you have the LS ground in the way suggested earlier then I’d have the ground lift in circuit which you could try carefully.

But then there's the oh no factor of if it isn’t broke don't fix it. What you could do is have it bypassed with a loop large enough to cut easily and then show Dave where it is.

Then if he has a hum problem with his kit then he can snip the link and have the ground lift in. TBH it really does depend on the grounding scheme for the preceding device. Pre or whatever.

Good work main man

Hi Chris

Yes I thought that those impedance figures where for the passive crossover and 8 ohms seems about right without it. Not having to drive the crossover should give you more sensitivity also as you won’t be burning up power through those resistors and inductors.

I heard those speakers passively driven and they sounded really good so active should push the quality up still further.

I know that we've had a chat about what you have in mind active Xover wise but others may be interested so any chance of a sneak peak at you plans.

But the potential of Alex's amps driving those speakers directly has really got my mouth watering.

Should be awesome

Take care

[Deleted]

1st, well done Sean!

Shaun, I tend to forget where I've said things, posted, pm'd, emailed whatever.

Anyhoosen;

The plan is,

Two dual mono stereo amplifiers (or maybe 4 mono blocks if I go case crazy) driving Naim SBLs directly (active). I was going to feed the amps from an Elliott active cross over but landed a Naim SNAXO 2-4 to make my life a little easier, I'll build the psu instead, I wonder if some JLHs might sneak their way inside the snaxo!

If you have never heard the SBLs active do try to search some out at a friends or something. Passive SBLs are merely decent at the original asking price, active they are simply stunning. As I can't afford the likes of ATC 50T or 100T ASLs, the SBLs really are the next best thing to my spoiled little shell-likes.

[Deleted]

Wooohoooo

Chris that sounds like heaven

Take care

[Deleted]

Apr 25, 2013 14:29:59 GMT sean said:

I replaced the output drive transistors, and so far it all seems OK now - output offset has come down to about 1mV (I can live with that!) under no load and about 0.1mV with a speaker load. Seems to play music well enough into the 4 Ohm speakers too. It's early days yet though, I need to get the second amp going now but that should be easy enough.

Hi Sean
That's great news. Many of us will be interested to hear what kind of a job it does driving those "difficult" speakers of yours as a pair of monoblocks, after you do the final adjustments when warmed up..
I hope that Dave is also going to throw in a carton of beer as well, for all the extra effort you needed to put into this build.

Now that the speaker ground is properly grounded directly, should I leave the ground lift resistor in place for signal ground or should I just link it out? It's linked out temporarily right now, and there are no problems at all with hum or anything. How was the 10R resistor causing problems originally out of interest, was it affecting the zobel network stability at all?

I would fit the 10 ohm earth lift resistors, as they may help to eliminate any low level hum with equipment that isn't connected to the same power board, or uses internal SMPS.
The best way that I could answer the other question is to quote what my E.E. friend Ionwyn from Sydney said when I asked him about the location of the Offset Corrector's input connection point.
I feel sure that Ionwyn will not mind me doing so. BTW, Ionwyn keeps the S.C. staff on their toes, and has recently mixed it with Douglas Self in a DIYAudio thread about Doug's preamp , although measuring impeccably, not sounding quite right to him. Yes, an E.E . that trusts his EARS as well as using test equipment.
Kind Regards
Alex

Taking the speaker return via 10R could result in either having a 6dB attenuator in the speaker output and a cooked 10R resistor or issues with the servo and higher distortion levels depending on the grounding topology......

[Will]

Hi Sean,

Good news on sorting the problem. Was it just the 327/337 you replaced?

[Deleted]

A progress report. The PCBs have yet to be double checked and tested. I didn't have the small heatsinks for the BC327 and BC337 fitted when the photo was taken.

Regards
Alex




Uploaded with ImageShack.us

[Deleted]

Hi Alex

Looking good

now just the fitting and testing to go

take care

[Deleted]

Hi Shaun
If I have made any errors during construction, it should be the same on both PCBs.
Regards
Alex
P.S.
They are nice looking PCBs, aren't they ?
Thanks Will !

[sean]

Apr 26, 2013 20:07:37 GMT Will said:

Hi Sean,

Good news on sorting the problem. Was it just the 327/337 you replaced?

Hi Will, yes it was - I sorted out the ground return for the speaker first of course, having the 10R resistor in the path must have caused the inital oscillations which then affected the 327/337 I think, which is why it still didn't work OK when the grounding was fixed. Looks like my DC corrector was working fine all along.

...I also changed the take-off point for the corrector from before to after the Zobel, might nto have made any useful difference but thought i might as well do it while I had the boards apart.