New amp testing, Noob needs input.

[buzz1167]

So I've built a Doug Self Blameless Amp, EF output, EF Vas. Using some of the transistors that were recommended to me from here,

ksa992FTA/ksc1845FTA for the input level
Ksc3503ESTU for the EF Vas
Ksc2690AYS and Ksa1220YSTU for the drivers
MJL4302A/4281A Outputs

I used .1ohm output resistors and I don't have a zobel network.

I've created some inputs for the amp and I'm trying to determine if it is properly stable or what I need to do to make it so. I got a program on the internet to output signals from my sound card like a signal generator, they aren't as good as a really nice generator, but I hope to tell something from them. Hopefully you guys can help...

I have run a complex sinewave - 1khz(Ref)+8khz(-15db)+16khz(-20db)
and a simple Sine wave at 30khz

I plotted an unloaded input (other channel) vs an unloaded output (No speaker) for each as well as an XY plot (hoping to show linearity/distortion)

What I get is this, For unloaded output the voltage values look really clean for both signals, and at 30khz the XY is a faint ellipse (showing some phase shift) probably <10deg. Plots are below...

Unfortunately for loaded output (just speaker, no zobel) the small signal response looks ok, but for any significant load the output voltage starts to "smear" on the oscope output, distortion/oscillation I assume... You can see this on the last picture. I don't know if its just me not knowing how to use the oscope correctly, or if its really something wrong with the output.

Is this just because I don't have a zobel? Maybe I need more outputs? But the voltage levels are still quite low < 10V pk. Basically I'm worried about packaging this setup because I don't want to finalize anything that's got a problem.

[davada]

Yes put the zobel in.
It looks like you've got a bit of cleaning up to do.

A schematic with all the component value posted would be necessary for anyone to help you with this.

Stick to a simple sine wave for now.

Is the sine wave you've shown both the input and output signal?
The traces are too thick possibly indicating some high frequency components, or it might just your test setup.

Post a pic of your layout as well.

[buzz1167]

Sorry for the awkward setup, but the layout on the schematic is representative of the layout on the board, I used it to place the components.
All the PNP transistors are Left flat and all the NPN's are Right flat. In reality I have changed the drivers and the VAS so they no longer look like a TO-92 package but I left the drawing as it was before.

[davada]

Hi Buzz1167,

I'm not used to seeing schematics drawn this way, but it's ok I can wrap my head around it.
I don't have a lot of time today to study this. I'll have a closer look at it later.

One thing that stands out to me is the omission of resistors between the driver transistor emitters and the output node. These resistor are necessary with an EF stage. Without them you can get a bit of high frequency oscillation occurring. The driver transistor emitter resistors are typically between 100 to 200 ohms. This is a good starting value. The end value depends on the amount of standing idle current the designer whats in the drivers. This depends partly on the transistor selection and the amount of bias current set in the output transistors.

I suggest adding the emitter resistors in and a zobel network as well. Be sure to place your zobel downstream from your feedback take off point.

If you are working the amplifier outside of a chassis, make sure you have the heat sink grounded. Any large surface area acts as an antenna for RF electrical noise in the environment.

Good luck with this,

David.

[Mooly]

Davada is correct... the drivers need that resistor. D Self cross couples the bases of the outputs with 200 ohm (It's not critical... 180, 220 all will do) and bypasses that with a 1uf poly cap to speed up the turn off of the outputs.

[buzz1167]

I've added a pic of what I went off of when I started the project.

I have tried to add that resistor (200R) between the bases of the output transistors. When I did I got a high frequency oscillation, observable with the oscilloscope at the shortest time interval (couple waveforms at .05us setting).
I also tried adding a .22uf cap across it as well which made the oscillations, unfortunately, more audible.

Also, I have multiple heat sinks, one for each power supply. So the heatsink is actually "powered". The goal was to attach the transistors per heat sink and have a "collector +" and "collector -" heat sink thus also sharing the power. I didn't previously know that grounding the heatsink was commonplace so I didn't purchase any electric insulating pads for the transistors.

[Mooly]

Hi Buzz,
That resistor/cap is vital... if it made matters worse you have other issues.

The problems you are experiencing (as said before in the other thread) will be down to layout and wiring details... and maybe the "live " heatsinks too.

I'm sorry we can't be more help... you must build it correctly. The amp is a fully worked example and it performs well. It needs no modifications or alterations. I built this amp on proper PCB's when it appeared 15 or so years back and used it for several years.

[Mooly]

Buzz... can you show us a photo of the amp... not saying it will help but we might spot something in your wiring.

When I first started building amps I went through all this, instability, oscillating, amps that were "nearly" right but that had weird problems such as oscillation under load on one side of the waveform. There's no magic fix... it's a steep learning curve, heck I'm still learning now after making it a career

[buzz1167]

Ive got a little better using the oscope so I got some better pictures, 30khz sine. Close up showing the oscillation everywhere with the 200ohm resistor.

The final pic that has less oscillation is simply with the 200ohm removed...

What would you like to see for the picture of the amp? I don't know whats going to help you. Since I don't have a prototype board I don't have a layout schematic, its just the same "hole board" I was using before.

To be honest, I can't really distinguish the oscillation without the 200ohm from audio, but with it, something does sound wrong in the high frequencies. Although hopefully it will just sound better all around when it stops this non-sense.

By the way, Listening to a 30khz signal for 5 or so minutes while setting this is really messes with your ears after you turn it off... It was weird...

Thanks again,

[davada]

Originally Posted by buzz1167

Also, I have multiple heat sinks, one for each power supply. So the heatsink is actually "powered". The goal was to attach the transistors per heat sink and have a "collector +" and "collector -" heat sink thus also sharing the power. I didn't previously know that grounding the heatsink was commonplace so I didn't purchase any electric insulating pads for the transistors.

Hi Buzz1167,

Don't do this. You have put two big antennas connected directly to the collectors of your output transistor. Your are injecting RF and other electrical noise directly into you power rails. This is asking for trouble.

RF likes to travel on the surface of things regardless if those things are grounded. But grounding dose attenuate RF greatly and prevents the RF from becoming an electrical current in a conductor. The best thing for RF interferance is is grounded shielding. However, even a transmitter of 5 Watts at close proximity can penetrate shielding.

Do invest in some insulators for the output transistors and the drivers should be on heat sinks as well. The EF darlington configuration is not known to be the most thermally stable arrangement.

I agree with Mooly, the emitters resistors are a must have and if putting them in makes things worse then the cause is something else. Have you put the zobel in?

You've got to get this oscillation out. It will severely effect the performance and stability of the amp.

David.