A missed opportunity ! I left my book at home .
Hi Ashok,
Nevertheless, I hope you can make it to RMAF or AES. I'll enjoy chatting with you. Maybe I can give you one of those Avery stickers signed so that you can affix it to your copy of the book
.Cheers,
Bob
Hello Bob,
I am testing my amplifier (revision 4) which is better than the other 3 (knowing 1 & 2 didn't work at all). But I still have one thing I can't seem to get rid of. I didn't find something in your book that could tell me if what I see is normal or abnormal.
I get near 50mV of noise on the scope trace. While tweaking and turning I can get it triggered and it seems there is a 20MHz wave in it (but not a real sinewave) but also much higher order rubble & noise. It is measured with no input and no load.
I have 4 grounds
1) input gnd
2) feedback gnd
3) common gnd
4) decoupling gnd
For now 1 and 2 are connected together with a little wire and go back to the power supply centertap.
Nulber 3 also goes back to the PS centertap (separate line). But there is also a resistor on the PCB wich is connected between CommonGround and Feedback/Input gnd. I often saw this and believe it's a good idea but maybe i am missing something here? It's a 10E resistor.
Number 4 goes to the PS centertap too (separate line)
There are no decoupling caps installed yet (I know N°4 is useless) But I have tried to place some caps w/o soldering them and it doesn't change anything at all (so I guess it is not decoupling cap related).
Maybe I should also say that this rubble is not only there at the output but more and less on every node (???). This makes me think nothing is wronf with the amp... it must be something else, something more external? Like GND issue? Scope Issue? Powersupply ?
Anyone any clues?
Thnx
Olivier
I am testing my amplifier (revision 4) which is better than the other 3 (knowing 1 & 2 didn't work at all). But I still have one thing I can't seem to get rid of. I didn't find something in your book that could tell me if what I see is normal or abnormal.
I get near 50mV of noise on the scope trace. While tweaking and turning I can get it triggered and it seems there is a 20MHz wave in it (but not a real sinewave) but also much higher order rubble & noise. It is measured with no input and no load.
I have 4 grounds
1) input gnd
2) feedback gnd
3) common gnd
4) decoupling gnd
For now 1 and 2 are connected together with a little wire and go back to the power supply centertap.
Nulber 3 also goes back to the PS centertap (separate line). But there is also a resistor on the PCB wich is connected between CommonGround and Feedback/Input gnd. I often saw this and believe it's a good idea but maybe i am missing something here? It's a 10E resistor.
Number 4 goes to the PS centertap too (separate line)
There are no decoupling caps installed yet (I know N°4 is useless) But I have tried to place some caps w/o soldering them and it doesn't change anything at all (so I guess it is not decoupling cap related).
Maybe I should also say that this rubble is not only there at the output but more and less on every node (???). This makes me think nothing is wronf with the amp... it must be something else, something more external? Like GND issue? Scope Issue? Powersupply ?
Anyone any clues?
Thnx
Olivier
Hi Olivier,
These kinds of things can result from many different things, and can be difficult to track down. There is a lot of RFI out there, and if you see the same stuff on the scope independent of node, it could be that kind of pickup. Do you see it with or withou the power to the amp on?
You seemed to say that the bypass caps are not in. This is hard to understand, as it would seem to wreak havoc with the circuits. It is really important to make sure that this is not the result of some local or global oscillations.
Best regards,
Bob
Hi Bob,
I did some checking and it seems the noise signal is somewhat everywhere even when the amp is off and even when the PSU is off. However the amplitude ranges from 10mV to 100mV depending on what is turned on or off or even depending on the node I measure (even outsied amp nodes). It also changes a bit when I alter the grounding wiring. Once the scope ground is on the signal ground, once the generator, once both, once it is on the PSU centertap etc... I still cannot figure out what it is exactly.
However I followed your lead for local oscillations (and indeed you mention it also in your book - measure more than just output). I stumbled uppon local oscillations that are somewhat 1V high in amplitude (!? funny they don't show up on the output and i also think it does not solve my noise issue -> but i must get rid of it anyway). One (for now), two in fact (upper side and lower side of the compl. amp) oscillations are located in the current source for the tailcurrent (?!) I use the constant current source of your book on page 40 fig 2.10f. But I added one resistor according to R. Slone's book. That is a base resistor of 1K in the Q2 base.
I searched there because I vaguely remembered Edmond Stuart said that resistor is bad and it seems he is right. Shorting it removes that oscillation. Now why is that resistor so bad? Can you say something about it? Meanwhile I search further to solve eventual other oscillations and my noise issue ...
Greetings
Olivier
I did some checking and it seems the noise signal is somewhat everywhere even when the amp is off and even when the PSU is off. However the amplitude ranges from 10mV to 100mV depending on what is turned on or off or even depending on the node I measure (even outsied amp nodes). It also changes a bit when I alter the grounding wiring. Once the scope ground is on the signal ground, once the generator, once both, once it is on the PSU centertap etc... I still cannot figure out what it is exactly.
However I followed your lead for local oscillations (and indeed you mention it also in your book - measure more than just output). I stumbled uppon local oscillations that are somewhat 1V high in amplitude (!? funny they don't show up on the output and i also think it does not solve my noise issue -> but i must get rid of it anyway). One (for now), two in fact (upper side and lower side of the compl. amp) oscillations are located in the current source for the tailcurrent (?!) I use the constant current source of your book on page 40 fig 2.10f. But I added one resistor according to R. Slone's book. That is a base resistor of 1K in the Q2 base.
I searched there because I vaguely remembered Edmond Stuart said that resistor is bad and it seems he is right. Shorting it removes that oscillation. Now why is that resistor so bad? Can you say something about it? Meanwhile I search further to solve eventual other oscillations and my noise issue ...
Greetings
Olivier
Hi Olivier, do you perhaps have fluorescent tube lights in your workroom? These can wreak havoc with scope measurements. They conduct as well as emit high levels of noise. If you have such a problem and you have two spare identical transformers make a proper line filter by connecting the primary of the first transformer to the mains, connect the two transformer secondaries together via a symmetrical Pi filter peaked at 50Hz and then use the output of the second transformer primary to feed your workbench power. This method cleans up a lot of conducted rubbish. If the two transformers are center tapped connect these together and run a clear earth from a ground peg outside. This is an essential piece of equipment for your electronic workshop.
For the record: Make sure that this real earth is an old car radiator or at least a meter long copper pipe buried in damp ground and it is the earth for your bench, not the mains earth at all.
For the record: Make sure that this real earth is an old car radiator or at least a meter long copper pipe buried in damp ground and it is the earth for your bench, not the mains earth at all.
Last edited:
Ground Loop ?
@ Olivier73
Hi, just a thought It's "possible" that you have a ground loop problem ? If more than one item are connected together are also mains earthed, that "could" be why. Try only having one item mains earthed & see if it makes a difference, & how etc
I'm NOT advocating disconnecting ANY of the chassis earths on the items, but rather ONLY having one signal earth connected to ground. This should usually be the driving signal ground item. eg: PreAmp or Signal Generator etc.
Regards
@ Olivier73
Hi, just a thought It's "possible" that you have a ground loop problem ? If more than one item are connected together are also mains earthed, that "could" be why. Try only having one item mains earthed & see if it makes a difference, & how etc
I'm NOT advocating disconnecting ANY of the chassis earths on the items, but rather ONLY having one signal earth connected to ground. This should usually be the driving signal ground item. eg: PreAmp or Signal Generator etc.
Regards
Hi Zero D, Andrew T & Nico,
Thnaks for the input. For now I cannot install a seperate ground lead.
My problem seems 2 ways.
1) I have local oscillations on the strangest places. I will make a printout and identify the nodes where I have oscillations. Placing caps helps but then again no... it's not always there ...it comes by touching the pcb and can go away by touching it elsewhere again ...
2) The noise on many nodes even powered off (but smaller). Here too I will make pictures and indicate on the schematic where it is measured.
Grrr this thing really gets me angry ... for now I take ZEN THEE before I start punching my pcb again (it tends to bite back)
I will post as soon as possible in the next few days (probably friday).
Cheers,
Olivier
Thnaks for the input. For now I cannot install a seperate ground lead.
My problem seems 2 ways.
1) I have local oscillations on the strangest places. I will make a printout and identify the nodes where I have oscillations. Placing caps helps but then again no... it's not always there ...it comes by touching the pcb and can go away by touching it elsewhere again ...
2) The noise on many nodes even powered off (but smaller). Here too I will make pictures and indicate on the schematic where it is measured.
Grrr this thing really gets me angry ... for now I take ZEN THEE before I start punching my pcb again (it tends to bite back)
I will post as soon as possible in the next few days (probably friday).
Cheers,
Olivier
In my lab, the battery charger for my portable drill generates massive amounts of RFI. I have to unplug it when doing any measurements or even listening to the shop's FM radio. The charger is so bad, that it even radiates to my neighbor's house and can be picked up by HIS radio. So maybe one of your neighbors has one of these...?
Not charging: Over an AM radio it sounds like "Braaap....Braaap.....Braaap...." A 60Hz square wave about once per 1.5 seconds.
Charging: The noise is almost constant "Braaaaaaaap...ip..ip..ip...Braaaaaap" as the power supply ramps the current up and down.
On FM - it detunes the station. Worth a look.
Hi everyone,
I decoupled all safety ground from scope, psu, generator.
It seems a bit better but it still is freaky.
It seems also however I suffer from something that was not when safety ground was connected : 50Hz Hum. Around 80mVpp.
However besides still struggling with noise I definately have a local oscillation problem. What kills me most is the fact it goes away just by touching some nodes or even just by ticking the PCB. This is very anoying since it is very hard to find out what betters or worsens the situation when you try something.
As I said I will (tomorrow or friday) post a schematic where I will indicate the local oscillation.
For the rest the amp works fine.
1,2Hz to 750KHz (-3dB). No peaking. Very stable. No overshoot. Waves look clean (on a 10Vpp / division scale). However waveform on a scope is not accurate and I don't have a THD analyser. But a clean looking square wave is a good start right? SlewRate = 100V/us.
Bye BYE for now
Olivier
I decoupled all safety ground from scope, psu, generator.
It seems a bit better but it still is freaky.
It seems also however I suffer from something that was not when safety ground was connected : 50Hz Hum. Around 80mVpp.
However besides still struggling with noise I definately have a local oscillation problem. What kills me most is the fact it goes away just by touching some nodes or even just by ticking the PCB. This is very anoying since it is very hard to find out what betters or worsens the situation when you try something.
As I said I will (tomorrow or friday) post a schematic where I will indicate the local oscillation.
For the rest the amp works fine.
1,2Hz to 750KHz (-3dB). No peaking. Very stable. No overshoot. Waves look clean (on a 10Vpp / division scale). However waveform on a scope is not accurate and I don't have a THD analyser. But a clean looking square wave is a good start right? SlewRate = 100V/us.
Bye BYE for now
Olivier
Uh. Wow. I think we found part of the problem.
I looked at your PCB layout. 4 layers is a blessing and a curse. Your efforts to make everything beautifully symmetrical are thwarted by a rather chopped-up ground plane with some isolation channels that could be forming unintentional tank circuits. Combined with your extraordinarily high bandwidth, I would be suspect of the layout, first.
See if bypassing the 10 ohm "isolation" resistor R0 with a 1nF X7R capacitor to increase coupling at RF will help you. Without a spectrum analyzer, you will have to find the magic value empirically, but 1nF is a good start.
Last edited:
Thanks, Bob. I'm afraid I don't have much else to contribute to the thread's subject. I just happen to subscribe to the thread and thought I could help as I have had some experience with managing 'weird' oscillations.
I can vow for this completely
You can even perform tricks to create and exploit those advantages. Guidance Troubleshooting
Hi there,
Here are some of my findings.
In the PDF of the schematic you can see some marked nodes RED and BLUE. The RED tracks represent the tracks/nodes where I have a 3,5Vpp 6,75MHz oscillation. The BLUE shows an oscillation of 50mVpp 31MHz.
All the other nodes are difficult to represent but here is in a nutshell : Nearly all nodes have a 10mVpp noise signal. Some also have (more or less clearly) a 15mVpp 520KHz signal superposed. On the output of the amplifier (once between the VAS and Dummy OPS + after the OPS) the same noise signal is there but at 80mVpp instead of 15mVpp which is almost only due to the 520KHz signal (that comes from I don't know).
As a matter of fact the 10mVpp noise signal + 520KHz shows up on the Common Ground connection too so it's logical this signal is everywhere.
I need to tell that the Feedback and Input Ground are together tied to the generator and scope ground which are protection grounded.
The PSU however is disconnected from ground. What I call common ground comes from the center tap of the PSU and also decoupling caps are routed to the centertap of the PSU. So only FB and IP ground are referenced to the real ground. I also have a connection between the real ground (IP+FB) and the PSU Centertap (DECOUPL GND & COMMON GND) through a 22E resistor on the PCB (R0).
I have two problems : 1) Oscillations 2) Noise
Problem is (1) can be : pcb layout related, RF pick up, local FB instability, ground loops, (2) can be : RF pick up, dirty common ground, ground loop
It can even be virtual due to the Node probing itself etc...
So the problems can be interconnected too ...
And most annoying is that it goes away sometimes ! (the oscillations , not the noise).
For example : the 6,75MHz 3,5Vpp oscillation which is very clear on the scope goes away just by touching with one 1 finger 1 of the miller caps CM1 or CM2 on the side that is connected to the input stage. Is that a lead i should invesitgate or rather coincidence?
Hard to tell what is the real offender here ... !?
You guy know more now? A breakthrough would really make my day !
Cheers
Olivier
Hi there,
Here are some of my findings.
In the PDF of the schematic you can see some marked nodes RED and BLUE. The RED tracks represent the tracks/nodes where I have a 3,5Vpp 6,75MHz oscillation. The BLUE shows an oscillation of 50mVpp 31MHz.
All the other nodes are difficult to represent but here is in a nutshell : Nearly all nodes have a 10mVpp noise signal. Some also have (more or less clearly) a 15mVpp 520KHz signal superposed. On the output of the amplifier (once between the VAS and Dummy OPS + after the OPS) the same noise signal is there but at 80mVpp instead of 15mVpp which is almost only due to the 520KHz signal (that comes from I don't know).
As a matter of fact the 10mVpp noise signal + 520KHz shows up on the Common Ground connection too so it's logical this signal is everywhere.
I need to tell that the Feedback and Input Ground are together tied to the generator and scope ground which are protection grounded.
The PSU however is disconnected from ground. What I call common ground comes from the center tap of the PSU and also decoupling caps are routed to the centertap of the PSU. So only FB and IP ground are referenced to the real ground. I also have a connection between the real ground (IP+FB) and the PSU Centertap (DECOUPL GND & COMMON GND) through a 22E resistor on the PCB (R0).
I have two problems : 1) Oscillations 2) Noise
Problem is (1) can be : pcb layout related, RF pick up, local FB instability, ground loops, (2) can be : RF pick up, dirty common ground, ground loop
It can even be virtual due to the Node probing itself etc...
So the problems can be interconnected too ...
And most annoying is that it goes away sometimes ! (the oscillations , not the noise).
For example : the 6,75MHz 3,5Vpp oscillation which is very clear on the scope goes away just by touching with one 1 finger 1 of the miller caps CM1 or CM2 on the side that is connected to the input stage. Is that a lead i should invesitgate or rather coincidence?
Hard to tell what is the real offender here ... !?
You guy know more now? A breakthrough would really make my day !
Cheers
Olivier
I'll risk first, although I'm not quite sure what Q15, Q16 are supposed to do...
The signals at their respective base and collectors are in phase, but of course Q15 and Q16 'whish' the opposite...hence the oscillations. First see what happens if you remove these two from the circuit.
I notice also Q25, Q26 as possible troublemakers, what is the DC voltage at their base?
The signals at their respective base and collectors are in phase, but of course Q15 and Q16 'whish' the opposite...hence the oscillations. First see what happens if you remove these two from the circuit.
I notice also Q25, Q26 as possible troublemakers, what is the DC voltage at their base?
Last edited:
I'll risk first, although I'm not quite sure what Q15, Q16 are supposed to do...
The signals at their respective base and collectors are in phase, but of course Q15 and Q16 'whish' the opposite...hence the oscillations. First see what happens if you remove these two from the circuit.
Q15 & Q16 are part of the common mode control loop (CMCL) that defines the VAS standing current. Without a CMCL this current can be anything between zero and a destructive huge value.
edit: Q25 & Q26 prevent more serious trouble.
Last edited: