PCB layout problems - Page 2 - diyAudio
Go Back   Home > Forums > Amplifiers > Solid State

Solid State Talk all about solid state amplification.

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 9th October 2012, 05:06 AM   #11
RJM1 is offline RJM1  United States
diyAudio Member
 
Join Date: Jan 2010
Location: Titusville, Fl.
akis, just open the schematic or view the pc board layers, hit the PrtScr key, open MS Paint, click on edit and then click paste. Save as any format you like.
  Reply With Quote
Old 9th October 2012, 07:57 AM   #12
akis is offline akis  United Kingdom
diyAudio Member
 
Join Date: Jul 2008
In the beginning of this thread I tried to show only the oscillator and the output stage and avoid other parts of the circuit which I have not even built on the PCB and would add confusion. I also could not get a jpeg file output, but now I have done it and am attaching it here.

To summarise in order to stabilise the circuit and eliminate harmonics I have needed to :

a) add big feedback caps to the driver transistors (600pF). These are not shown on the jpeg schematic. On the breadboard model, these were sufficient at 100pF whereas on the PCB I needed 600pF.

b) increase the bias of the output stage until the curve does not suffer from crossover distortion but also until it looks cleaner on the scope - typically 100mA idle. There is no difference here between PCB and breadboard

c) add a zobel of 1R/330nF at the output to eliminate parasitics at the output when the load is a transformer - again no difference between PCB and breadboard

d) add a sheet of aluminium under the PCB and connect this to the circuit "common" - no difference between PCB and breadboard as the breadboard is mounted on an aluminium plate.

e) the wien bridge oscillator on the breadboard is "steady" and not affected by the following stages, scope probes etc. For example I can connect it or disconnect it from the output stage with only a minimal change in waveform. On the PCB however it seems it is sensitive to everything that follows and much harder to control. The 600pF feedback caps have stopped the unwanted oscillations but I am not sure what else is lurking there. I was expecting that the output of U1.2 should be rock steady and imprevious to almost anything that follows short of a short This is true on the breadboard but absolutely not true on the PCB.

I am thinking that my PCB design is not proper, I have seen other PCBs with huge ground planes on both sides, which is not something I have done (or know how to do). But it may just be something else I have missed. If you cannot see the PCB from the gerber files then I will try to post jpegs of it, even though it is two layered.

Edit: managed to get a better jpeg
Attached Images
File Type: jpg version1.jpg (115.3 KB, 91 views)

Last edited by akis; 9th October 2012 at 08:08 AM.
  Reply With Quote
Old 9th October 2012, 07:58 AM   #13
akis is offline akis  United Kingdom
diyAudio Member
 
Join Date: Jul 2008
Quote:
Originally Posted by RJM1 View Post
akis, just open the schematic or view the pc board layers, hit the PrtScr key, open MS Paint, click on edit and then click paste. Save as any format you like.
And would you believe that that is what I do most of the time, except it does not work for Diptrace... I have now done it using "print preview"...
  Reply With Quote
Old 9th October 2012, 08:15 AM   #14
akis is offline akis  United Kingdom
diyAudio Member
 
Join Date: Jul 2008
Quote:
Originally Posted by Dave Zan View Post
You probably have a couple of problems.
That can lead to confusion if they interact.
1. It is essential to decouple the power supply to the WB Oscillator section.
Had practical experience with that.
2. The CFP output that you use is notorious for instability.
If you are not already aware of the issues then there are multiple threads to search. I have just started to study this because fixes seem to be more trial and error than I trust.
Avoid the problem with EF unless you really need CFP.
If so then you can try base stoppers on the outputs, the drivers, between the output collectors and the driver emitters.

Best wishes
David


BTW. Hard to keep everyone happy but many people don't like JPGs either and prefer a lossless format.

1) I have done that from the start, as you can see from post #12 there are 120R resistors followed by 100uF/100nF caps very close to the op-amp. On the breadboard I can remove completely the resistors and bypass caps and it makes no difference whatsoever to the quality of the output, but I left them in place just in case.

2) I do not understand what is "CFP output" ? What is "EF" ? What is "base stoppers" ?
  Reply With Quote
Old 10th October 2012, 12:18 AM   #15
diyAudio Member
 
Join Date: Nov 2010
Quote:
Originally Posted by akis View Post
I do not understand what is "CFP output" ? What is "EF" ? What is "base stoppers" ?
There are 2 common ways to connect the output transistors.
EF = Emitter Follower - The default "standard".
CFP = Complementary Feedback Pair - Some potential benefits but more prone to oscillation. Your circuit is CFP and this is very probably one of your problems.
D. Self and Bob Cordell both discuss these options and provide essential information on the fundamentals. After that there are several, easy to find threads here on CFP stability problems and preventative measures - one of which is a resistor in series with the base, called a[n oscillation] "stopper". Your circuit lacks these so it is not a surprise that it is borderline stable.

Best wishes
David
  Reply With Quote
Old 10th October 2012, 01:11 AM   #16
diyAudio Member
 
nigelwright7557's Avatar
 
Join Date: Apr 2008
Location: Carlisle, England
I created a pcb for a mixer without putting any thought into ground loops.
The pcb came back, I built it up and it hummed like mad. I tried shorting the ground lines with wires but this had limited success.

I redesigned the pcb seperating the audio ground and power supply ground and it worked a treat. There musnt be any power supply currents in the audio ground line so a single point for the grounds to meet is vital.

For class d I found that I needed to keep tracks as short as possible with decoupling as close to the class d chip as possible.
__________________
PCBCAD50 software. http://www.murtonpikesystems.co.uk
  Reply With Quote
Old 10th October 2012, 07:44 AM   #17
akis is offline akis  United Kingdom
diyAudio Member
 
Join Date: Jul 2008
Quote:
Originally Posted by Dave Zan View Post
There are 2 common ways to connect the output transistors.
EF = Emitter Follower - The default "standard".
CFP = Complementary Feedback Pair - Some potential benefits but more prone to oscillation. Your circuit is CFP and this is very probably one of your problems.
D. Self and Bob Cordell both discuss these options and provide essential information on the fundamentals. After that there are several, easy to find threads here on CFP stability problems and preventative measures - one of which is a resistor in series with the base, called a[n oscillation] "stopper". Your circuit lacks these so it is not a surprise that it is borderline stable.

Best wishes
David
That is very helpful, thank you. I have done a search and will continue looking out for it.
  Reply With Quote
Old 10th October 2012, 08:15 AM   #18
diyAudio Member
 
Join Date: Nov 2010
It's a pleasure. Perhaps it is too late now that you have a PCB but I would recommend the less troublesome EF unless you have a specific requirement for a CFP. That doesn't seem to be the case here.
There is not much detailed analysis of the CFP so if you continue with it and learn any new lessons then be sure to post them.

Best wishes
David

Last edited by Dave Zan; 10th October 2012 at 08:34 AM.
  Reply With Quote
Old 10th October 2012, 09:40 AM   #19
akis is offline akis  United Kingdom
diyAudio Member
 
Join Date: Jul 2008
I have preferred the compound pair rather than the darlington based on:

1) the darlington has double the voltage drop from the rails. If you are working with batteries then losing 1.2 V from the rails is very significant. With CFP I lose just 600mV.

2) based on Rod Elliot's research the compound pair is much more thermally stable than the darlington pair and he prefers it in all his designs. A link: Elliott Sound Products - Audio Power Amplifier Design Guidelines

So far I have discovered:

I have noticed that the CFP's driver transistor is in common emitter mode, ie it has a resistor on its collector and nothing on the emitter (no feedback resistor), therefore, as seen in isolation from whatever follows, we have a 82R-120R resistor on the collector and a 8R load on the emitter, giving us a voltage gain of 10. Voltage gain usually means trouble, whereas a common collector arrangement should not suffer because there is no voltage gain. So I added the B to C caps to stem any voltage gain at HF, which on my scope looked like 10X-20X the fundamental, ie around 2MHz-4MHz.

On the breadboard there are contact resistances which alter the behaviour of a circuit when large currents pass through (> 50mA). For example I was building a Li-Ion battery charger on breadboard and could not understand the strange voltage readings. It was because large currents of >50mA were causing errors large enough to disturb the rest of the circuit. To get around this problem I identified which component's pins would be passing more than 50mA and soldered two wires on them so that they subsequently connected to the breadboard with two contacts for each one pin. This made things much better, but still not as good as soldered connections.

I also understand that with the CFP this is the same case on breadboard (CFP testing usually means >50mA), so when it comes to HF oscillations the breadboard behaves way differently to the PCB. This must be taken into account.

From the three proposed solutions, ie resistors at bases, resistors at emitters and caps on B to C, I think the latter is the best because you do not lose any voltage from the rails. For example with 82R on the collector and 10R on the emitter as a "HF stopper", you already lose 100mV-200mV - and am not sure if the 10R will stop the HF oscillations at all.

In my case my fundamental frequency is 200KHz, 10 times the audio upper limit, and the 600pF caps have no effect on the output voltage swing, so I suspect that they would also not have any effect on audio frequencies.

I have also found that driving an inductive load (eg a transformer) is very different to driving a resistive load (eg a dummy load resistor). But I added a zobel of 1R/18nF and that stopped oscillations completely.

Biasing is also important, low bias creates HF oscillations when driving a heavy load, you need to increase the bias while under heavy load conditions until the curve looks right on the scope. In my case with +/-8V at the output and a 8R load, I needed about 100mA on "idle" to get a good looking curve at full load.
  Reply With Quote
Old 10th October 2012, 11:15 AM   #20
diyAudio Member
 
Join Date: Nov 2010
Quote:
Originally Posted by akis View Post
I have preferred the compound pair rather than the darlington based on:

1) the darlington has double the voltage drop from the rails. If you are working with batteries then losing 1.2 V from the rails is very significant. With CFP I lose just 600mV.

2) based on Rod Elliot's research the compound pair is much more thermally stable than the darlington pair and he prefers it in all his designs.
You trade a major risk of electrical instability for thermal stability improvement that is more theoretical than useful because it is not hard to thermally stabilize an EF anyway.
I have read Rod's articles but it seems an unfavourable trade-off unless you really need the extra 600 mV or have some exceptional case.
Makes some sense for ultra low distortion class A and that's what piqued my own interest.
What's your application? Looks like a Sonar driver or some ultrasonic transducer.
In your case the particular stray capacitances and resistances of the breadboard seemed to have helped to stabilize that layout. Too accidental for my comfort zone, I want a clear analysis, but best of luck.

Best wishes
David
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Eagle vs. Sprint-Layout for PCB design/layout hollowman Parts 11 12th January 2014 09:01 PM
Matchless Hotbox Layout problems bagudan Instruments and Amps 9 17th July 2012 06:56 PM
Please check my PS pcb board layout for problems???? marchewd Tubes / Valves 6 10th July 2008 05:47 PM
preamp layout and hum problems - help! Tex Houston Solid State 1 11th June 2007 09:44 AM
Pcb Layout KIERANPHILLIPS Solid State 1 19th March 2002 02:49 PM


New To Site? Need Help?

All times are GMT. The time now is 11:53 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2