OK, Sander,
Thanks for the dialogue, this is all good - could you start with a careful tabulation of the parts? Shouldn't take too long, put the parts issues to rest - then inspect a couple of Lgreen's photos. Do you have any pictures of local amps that are known to work well?
I have to say that from my own design and construction experience, if the amp is utterly quiet, there is no smoke, offsets and bias levels are all correct as they are here, and pcbs have been supplied, then issues with oscillation and poor square wave performance come back to design, or dimensioning, or both. This is not to criticise, but it has also been the experience of many others.
BTW, you should not assume that Lgreen is an amateur......
Thank you for cooperating,
Hugh
Thanks for the dialogue, this is all good - could you start with a careful tabulation of the parts? Shouldn't take too long, put the parts issues to rest - then inspect a couple of Lgreen's photos. Do you have any pictures of local amps that are known to work well?
I have to say that from my own design and construction experience, if the amp is utterly quiet, there is no smoke, offsets and bias levels are all correct as they are here, and pcbs have been supplied, then issues with oscillation and poor square wave performance come back to design, or dimensioning, or both. This is not to criticise, but it has also been the experience of many others.
BTW, you should not assume that Lgreen is an amateur......
Thank you for cooperating,
Hugh
Last edited:
Hugh,
Well, looking at the pictures on his website he seems a guy that's quite happy to build a few amplifiers. Whether he understands what goes and what doesn't is another, wire spaghetti seems to be his forté. But all that can be dealt with with some advice and rework I guess.
Cheers,
Sander.
Well, looking at the pictures on his website he seems a guy that's quite happy to build a few amplifiers. Whether he understands what goes and what doesn't is another, wire spaghetti seems to be his forté. But all that can be dealt with with some advice and rework I guess.
Cheers,
Sander.
Attached are two PDF documents detailing component placement on the amplifier and the power supply PCB. A previous version of these documents has been sent to everyone who purchased a set of PCBs but I updated them so even a blind monkey wouldn't be able to stuff the wrong part in the wrong location.
Cheers,
Sander.
Cheers,
Sander.
Thank you, Sander.
Doubtless all this is elsewhere in the thread, but I'm sure I speak for all when I say its updated reappearance here and your desire to help is much appreciated.
Cheers,
Hugh
Doubtless all this is elsewhere in the thread, but I'm sure I speak for all when I say its updated reappearance here and your desire to help is much appreciated.
Cheers,
Hugh
Well, I do not post often, but I am also in the process of building a stereo version of ExtremeA, actually I will have 2 of them eventually. I am not at the stage where I can contribute to the verification of voltages at various nodes, but I must echo some of the comments made by Sander regarding the component selection and particularly wiring.
I had in the past worked for Bryston, in an engineering capacity where I had the chance to see what happens if even an inch of wire added against the advice of someone who had a direct experience with an amplifier. I had Bryston amps literally showing a THD+N behaviour from 0.05% all the way to near 1% just because the wire emanating from the feedback network to ground was either too long or improperly twisted. Go figure! This is just an example of what might happen and not a statement on a LGreen’s build.
Another thing, - it would worry me if the amp is running as hot as Lgreen built it. In my experience going above 50-55 deg. C sometimes brings an unknown that is nearly impossible to pin down. I had this issue with my attempt to build A75 of Pass/Thagard fame many years ago. Sander’s suggestion of going down to 20W is what I would do. I bet that the oscillations would be gone when the power is turned down.
I am doing a DIY heatsink that will most likely have 6-inch long fins (L-brackets bolted to the 0.75 inch thick aluminum plate 13.5 inches long and 7 inches tall). Except for the fins it is similar to what Lgreen used. This is the minimum for class-A amp doing about 150W dissipation (75W) according the simulations I did on R-theta site with their heat-sink simulator. Smaller fins, as in Lgreen build, would lead, in my view, to an excessive heat and that might make an amp to become marginally stable at certain frequencies, - an utterly unpredictable behaviour.
I have some amplifier design experience and after staring at the schematics for nearly 4 months, in my view, ExtremA should be, as Sander says, - stable. I do not see any issues here, - all transistors used in cascodes along with current source/sinks are operating in the appropriate regime. I also did some simulations last summer using a rather cheesy Multisim software (ver.10). Although, the output transistors models were primitive, but still, I do not recall seeing any issues. Clearly the amp should work.
regards,
Vadim
I had in the past worked for Bryston, in an engineering capacity where I had the chance to see what happens if even an inch of wire added against the advice of someone who had a direct experience with an amplifier. I had Bryston amps literally showing a THD+N behaviour from 0.05% all the way to near 1% just because the wire emanating from the feedback network to ground was either too long or improperly twisted. Go figure! This is just an example of what might happen and not a statement on a LGreen’s build.
Another thing, - it would worry me if the amp is running as hot as Lgreen built it. In my experience going above 50-55 deg. C sometimes brings an unknown that is nearly impossible to pin down. I had this issue with my attempt to build A75 of Pass/Thagard fame many years ago. Sander’s suggestion of going down to 20W is what I would do. I bet that the oscillations would be gone when the power is turned down.
I am doing a DIY heatsink that will most likely have 6-inch long fins (L-brackets bolted to the 0.75 inch thick aluminum plate 13.5 inches long and 7 inches tall). Except for the fins it is similar to what Lgreen used. This is the minimum for class-A amp doing about 150W dissipation (75W) according the simulations I did on R-theta site with their heat-sink simulator. Smaller fins, as in Lgreen build, would lead, in my view, to an excessive heat and that might make an amp to become marginally stable at certain frequencies, - an utterly unpredictable behaviour.
I have some amplifier design experience and after staring at the schematics for nearly 4 months, in my view, ExtremA should be, as Sander says, - stable. I do not see any issues here, - all transistors used in cascodes along with current source/sinks are operating in the appropriate regime. I also did some simulations last summer using a rather cheesy Multisim software (ver.10). Although, the output transistors models were primitive, but still, I do not recall seeing any issues. Clearly the amp should work.
regards,
Vadim
I tried to look at the huge pictures where the star ground is done to the chassis but the pictures are so blurry it's difficult to see, and they are also way too big, takes long time to load up the picture and don't even fit on my 24" wide screen.
L-green I downloaded 2 pictures and could read out from the EXIF data that the pictures were taken with a nice Olympus E-510 DSLR system camera but the aperture is set to lowish F4, that's why the pictures have such a small DOF (Depth of Field), an aperture set to at least F11 or preferably even higher would be better but then the shutter speed goes down dramatically so it will for sure create hand shake blur so you should take picture in very good light or help up with aid of the flash light, enough camera talk. 🙂
Did you have any star ground made in the chassis?
Cheers Michael
L-green I downloaded 2 pictures and could read out from the EXIF data that the pictures were taken with a nice Olympus E-510 DSLR system camera but the aperture is set to lowish F4, that's why the pictures have such a small DOF (Depth of Field), an aperture set to at least F11 or preferably even higher would be better but then the shutter speed goes down dramatically so it will for sure create hand shake blur so you should take picture in very good light or help up with aid of the flash light, enough camera talk. 🙂
Did you have any star ground made in the chassis?
Cheers Michael
Last edited:
It is a little more difficult to measure this amplifier because of its differential nature and its happiness for differential signals, as my Tek FG is not differential. I feed the FG signals through a converter to get them balanced.
In post #17 we were talking about the square wave response, I said that my square waves were going into my SE->BAL converter and that could have been an issue.
Tonight I looked at this. But first, my scope channel #2 was acting up (as per Post #12 and Post #14) and I discovered that the probe was faulty (intermittent connection) so I put on a new probe and everthing works great. I am able to get differential measurements.
Here are images of INPUT signals.
Square waves-
INPUT 2 KHz from BAL converter into Amplifier
INPUT 20 KHz from BAL converter into Amplifier.
(Same result unloaded straight into scope).
So we see that the BAL converter is causing some ringing.
I cannot reliably push differential square waves into the amp.
So I connected the function generator directly to the RCA input on the ExA and flipped the handy dandy switch to ground the negative input.
Here is 20KHZ output of ExA into 4 ohms from the SE input. Not sure it matters (given the SE input), but this is the full differential output.
Both channels behave the same.
Perhaps we should not panic about the square wave response?
In post #17 we were talking about the square wave response, I said that my square waves were going into my SE->BAL converter and that could have been an issue.
Also mentioned in Post #12.
Tonight I looked at this. But first, my scope channel #2 was acting up (as per Post #12 and Post #14) and I discovered that the probe was faulty (intermittent connection) so I put on a new probe and everthing works great. I am able to get differential measurements.
Here are images of INPUT signals.
Square waves-
INPUT 2 KHz from BAL converter into Amplifier
An externally hosted image should be here but it was not working when we last tested it.
INPUT 20 KHz from BAL converter into Amplifier.
An externally hosted image should be here but it was not working when we last tested it.
(Same result unloaded straight into scope).
So we see that the BAL converter is causing some ringing.
I cannot reliably push differential square waves into the amp.
So I connected the function generator directly to the RCA input on the ExA and flipped the handy dandy switch to ground the negative input.
Here is 20KHZ output of ExA into 4 ohms from the SE input. Not sure it matters (given the SE input), but this is the full differential output.
An externally hosted image should be here but it was not working when we last tested it.
Both channels behave the same.
Perhaps we should not panic about the square wave response?
Last edited:
Thats much better.
As suggested in post #16, before all that excitement:
"Check the sig gen OP, each phase of unbal -> bal converter
unloaded, same loaded and then amp OP."
Time for a friday night beer.
An externally hosted image should be here but it was not working when we last tested it.
cheers
A good result, a nicely rounded square wave, exactly as I would expect in a good amp.
I guess I owe you an apology for doubting the amp, Sander, I am sorry. It looks good, could we also have a view of the clip, Lgreen?
Lgreen, differential measurements do take more care..... you might consider another probe!
Terry, almost thirty years ago, in 1980 when I was a young fit Army guy, I once went on a long, Hobart jog, and the halfway point was exactly that spot at the old Cascade Brewery!
Cheers,
Hugh
I guess I owe you an apology for doubting the amp, Sander, I am sorry. It looks good, could we also have a view of the clip, Lgreen?
Lgreen, differential measurements do take more care..... you might consider another probe!
Terry, almost thirty years ago, in 1980 when I was a young fit Army guy, I once went on a long, Hobart jog, and the halfway point was exactly that spot at the old Cascade Brewery!
Cheers,
Hugh
Cascade Premium
We toured the brewery about 2 months ago, very interesting.
I bet that was a tough day - you obviously remeber it well.
We did the group mountainbike ride from top Mt Wellington down past
brewery to hobart city. A lot of fun, especially for the kids.
OK, apologies for OT posts.
T
That looks better.
I had a feeling, despite the impossible, evil complexity of the amp 🙂rolleyes🙂, that it would perform better than pictured.
I really like this amp, the more I look at it, the more I like it. Could be a future project if I ever get my current projects finished.
I had a feeling, despite the impossible, evil complexity of the amp 🙂rolleyes🙂, that it would perform better than pictured.
I really like this amp, the more I look at it, the more I like it. Could be a future project if I ever get my current projects finished.
Onset of clipping
My wife switched camera lenses on me and I am a novice at DSLRs so sorry about the weird pictures. With the new lens everything is in your face.
Onset of clipping at 2KHz. When I looked at this the differential signal went off screen so here are zoom ins of just the + output.
You do see these same oscillations on the differential plot.
Onset of clipping 1
Onset of clipping 2
My wife switched camera lenses on me and I am a novice at DSLRs so sorry about the weird pictures. With the new lens everything is in your face.
Onset of clipping at 2KHz. When I looked at this the differential signal went off screen so here are zoom ins of just the + output.
You do see these same oscillations on the differential plot.
Onset of clipping 1
An externally hosted image should be here but it was not working when we last tested it.
Onset of clipping 2
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
Gain?
What is the voltage gain of the ExA?
It is possible for these clipping measurements (which did use the BAL converter) that the battery-operated DRV134s were clipping. I did not have time in the middle of the night to check this but its certainly possible and occurred to me this morning after posting these pics.
What is the voltage gain of the ExA?
It is possible for these clipping measurements (which did use the BAL converter) that the battery-operated DRV134s were clipping. I did not have time in the middle of the night to check this but its certainly possible and occurred to me this morning after posting these pics.
Anyway, I have seen that the oscillations at clipping do occur when this is driven directly from the FG (SE) so I don't think its my BAL converter. In such a case (driven SE) and looking at the different ouputs; the + output tends to osclliate first at its most negative value and the - output tends to osc. first at its most + value.
When does the forki lift truck arrive to lift it ? lol
I use my own design high power amps on my disco and they are very heavy.
I dont skimp on power transformers or heatsinks hence the extra weight.
I use my own design high power amps on my disco and they are very heavy.
I dont skimp on power transformers or heatsinks hence the extra weight.
I have seen similar oscilations on my amps at post design testing stage.
It was usually due to too much gain on the drivers.
I put 220pF caps between the base and collector on the drivers to fix it.
Another place that might be the problem is the c-b capacitor on the VAS.
It was usually due to too much gain on the drivers.
I put 220pF caps between the base and collector on the drivers to fix it.
Another place that might be the problem is the c-b capacitor on the VAS.
Thank you for a constructive idea. I appreciate the suggestion. Hopefully another build will be finished before I have to change anything. If not, this is easy to implement.
- Status
- Not open for further replies.