"Blameless" style ~100W amp with CFP output.

[mnturner]

Hey guys, this is my first real post in the SS forum.
Just got my "Blameless" Self/Sloan amp up and running and thought I'd post it. I was running a 2*50W Gainclones, but I felt like a new project and thought I'd build something with the potential for a bit more power. I haven't made any detailed measurements of distortion yet, but I will do so and post results if there's any interest. For what it's worth, it sounds great.
There's no DC servo, even without DC offset was about 9mV. It runs pretty quiet aside from some weird triangle wave shaped noise which I'm pretty certain was coming from the switcher in my PC which was sitting right next to the amp; posting a pic, if anyone has any other ideas about what it is please speak up. I won't be able to test properly for noise until I've built a sold power supply for it and can box it up properly with some shielding.
In the image, its running on 17V rails, with a small heatsink for testing puroposes.
I designed the single sided PCB layout with EagleCAD and etched it by hand. I'll post the layout if anyone is interested.

[alexcp]

If I am reading the settings of your scope correctly, the sawtooth wave has a period of 20mS, that is, its frequency is 50Hz. If that's the case, it's coming from your amp's power supply.

[Mooly]

Originally Posted by alexcp

If I am reading the settings of your scope correctly, the sawtooth wave has a period of 20mS, that is, its frequency is 50Hz. If that's the case, it's coming from your amp's power supply.

Agreed... and probably from input/feedback return grounding points. The output should have no identifiable 50/60hz component when the input is shorted to input ground, even with significant ripple on the rails.

[AndrewT]

Hi,
the input resistor, R5=10k, that defines Rin may not be exactly 10k.
It might reduce the output offset by lowering R5 slightly or raising R5 slightly.

It is easy to check this.
Let the amp warm up monitoring the output offset. When it is stable, add an extra resistor across R5 in parallel. try 100k. Does the output offset increase or decrease?

If the output offset decreases trying adding a lower value resistor, try 47k. Keep going till you find the value that gives the lowest offset. I doubt you will need to go lower than 8K (=10k//39k).

Alternatively if making R5 lower increases output offset then you can try increasing R5.
This is not so easy to keep re-trying by trial and error.
Instead remove 10k and fit 12k. Measure the fully warmed up output offset. Has the output offset increased or reduced. try adding 100k in parallel. Is output offset higher or lower.

I am not doing this to minimise output offset which it does very effectively. I am suggesting this in the hope that you can increase Rin above the 10k value that you have at present. This is very low by domestic standards and limits the usability of the amplifier. You might find that 10k5 gives zero offset. or 11k2 or some other value.

If a lower than 10K Rin gives lowest output offset, you can consider swapping T1 and T2
to find the new Rin for lowest offset.

BTW,
bc556a does not make a good input transistor. The low gain results in high input offset current and this gives high voltage across Rin and also leads to higher than needed output offset. Try to find equivalent transistors with gain >400. Some types go over 1000 hFE, but they are more difficult to source and generally much more expensive.
bc556c and bc560c are generally in the range 400 to 600 hFE.
T10 and T8 look like they are running very warm. Attach flag cooling or replace with >=1W devices.

[wintermute]

A few more details about how the measurement with the cro was done would be good It is the output of the amp that it is showing right?

The fact that it is 50 Hz is interesting in itself (to me) I'm assuming you have a full wave rectifier, normally you will get 100Hz ripple if you have full wave rectification. Assuming a dual secondary transformer, If the secondary's are wired out of phase though you will get 50 Hz ripple. You can check this by measuring the AC voltage across the bridge, if it is zero volts then the secondary's are wired out of phase.

The other thing that is interesting is how clean the waveform looks, apart from the 50Hz sawtooth there doesn't appear to be any other "noise" at all! It looks more like the output of a well filtered power-supply to me!

Tony.

[mnturner]

Thanks for the feedback guys.
The 50hz hum might be the result of the power supply I had it running off at the time; it was a poorly built and badly laid out PSU I built a while ago for something else with a transformer I'd pulled out of a broken amp I came across. The grounding on the PSU board itself is poor and as suggested the trafo winding may be in the wrong phase. (Yeah, that was the output of the amp Tony.)
Andrew T, the lower than usual input impedance is recommended by D. Self to, if I recall correctly, reduce the amound of interference/noise the input stage picks up by using lower Z nodes. I'm probably going to run it with a preamp that I'll make sure is punchy enough to drive the lowish input impedance.
The input transistors are BC546C's, I agree that higher beta would be better though. I pretty much just used them because I wasn't sure how high I'd end up making the rail voltages, so I used high voltage transistors throughout.
Yeah T10 and T8 are running kinda close to maximum spec, I'll see how they go.
In regards to reducing the output offset, I probably won't bother to try, 9mV really doesn't worry me.

[AndrewT]

bc556c have far more gain than bc556a.
c grade are OK for input pair.

[mnturner]

Hooked it up to a good 35-0-35v power supply with 10,000uF per rail and its running sweet. The hum seems to be totally gone, must have been a grounding issue.

[Mooly]

That's good, and heed the warning re T8 and T10 or it might start humming very loudly...

[AndrewT]

I don't think he understands that To92 devices must be temperature de-rated just like power devices when Tc > 25degC.