I was not going to reply but Angle has some valid points.
TM. Transient Miller Compensation. An amplifier topology in the second stage of a conventional three stage amplifier controlling the phase shift in the feedback to maintain stability.
DP. Dominant pole filtering. Just using a filter after the second stage. Increases stability by filtering the high frequencies so they can't feed back thereby limiting the phase shift of the feedback. In very simple terms.
Doubling the outputs. Well, get a transistor data book and look at the curves. Understand them. Pay careful attention to the thermal factors.
I suggested reading the design book by Robert Self. He describes the design of amplifiers in great detail. If you are going to mod something, you had better know how they work. Then you can decide if you know better. A book is a lot cheaper than sets of outputs. I know this first hand.
I also suggested reading the viewpoints from Nelson Pass. He has a different view on what sounds good from either Self or Curl. Might even dig up the papers from Leach, Bolbery and Didden. I am sure there are others. The AA series POOGE tells you a lot about trade-offs made for mass production and how to improve them from a more hobbyist viewpoint.
I have used LT Spice both for hobby and professionally for decades and it has identified numerous design decisions I do not agree with as well as plain old mistakes. Mistakes made by very smart professionals. Realize, many of the SOPs were established and taken for granted before we had such tools. Do a full montecarlo analysis and you may be surprised. To just build what the model says is just as ignorant as not doing the analysis. To include all of the parasitic factors is darn near impossible unless you have total control, like within an IC. What was the old phrase? "Trust but verify"?
TM. Transient Miller Compensation. An amplifier topology in the second stage of a conventional three stage amplifier controlling the phase shift in the feedback to maintain stability.
DP. Dominant pole filtering. Just using a filter after the second stage. Increases stability by filtering the high frequencies so they can't feed back thereby limiting the phase shift of the feedback. In very simple terms.
Doubling the outputs. Well, get a transistor data book and look at the curves. Understand them. Pay careful attention to the thermal factors.
I suggested reading the design book by Robert Self. He describes the design of amplifiers in great detail. If you are going to mod something, you had better know how they work. Then you can decide if you know better. A book is a lot cheaper than sets of outputs. I know this first hand.
I also suggested reading the viewpoints from Nelson Pass. He has a different view on what sounds good from either Self or Curl. Might even dig up the papers from Leach, Bolbery and Didden. I am sure there are others. The AA series POOGE tells you a lot about trade-offs made for mass production and how to improve them from a more hobbyist viewpoint.
I have used LT Spice both for hobby and professionally for decades and it has identified numerous design decisions I do not agree with as well as plain old mistakes. Mistakes made by very smart professionals. Realize, many of the SOPs were established and taken for granted before we had such tools. Do a full montecarlo analysis and you may be surprised. To just build what the model says is just as ignorant as not doing the analysis. To include all of the parasitic factors is darn near impossible unless you have total control, like within an IC. What was the old phrase? "Trust but verify"?
Oh no, please don't start winding each other up again.........
Get some Zen in and chill to the positive Chi from the east
tvrgeek (is that a reference to the great the sportscar I think it is?) did put in an effort to explain things which is much appreciated.
I am not familiar with the term Transient Miller Compensation (TM) - is that akin to the two-pole Miller I tried 5 years ago in post #105 on this thread? Or is it what Douglas Self calls output-inclusive Miller compensation?
Which I had to reverse to a standard feedback Miller after getting oscillations. At the time I concluded that it was due to the hand taped RA-820AX layout and component tolerances. No Spice simulation could have solved this, as tvr also rightly points out.
Then, there are different types of Dominant Pole Compensations (DP). Stan Curtis who designed most, if not all the Rotels always used a DP shunt topology, loading the VAS transistor collector with 33k in parallel with 330pF to ground. (Plenty schematics showing this at the start of this thread)
This heavy handling of the VAS is what produces the round/mellow "Rotel sound". It definitely stabilizes the amp, reducing after-sales service costs - but actually many people just loved it - some still swear by this sound.
I didn't mind it until Chris (post #119) tempted me to go for some 'fun and giggles' - which I did. And I have never looked back, nor has e.g. Paulus in his RB-991 upgrades (see post #394).
dreamth on the other hand suggested (at least I think he did) that doubling the drivers would be great. Yes, that will help the output stage drive, but wouldn't it also double the (non-linear) capacitive and current load on the previous VAS stage? What can/should be done about that - except for adding an extra driver stage? Also, to what level should these paralleled drivers be matched to ensure that one is not doing all the work while its partner is just freewheeling? Does anyone know if there are any integrated dual matched chip driver transistors - as there are internally chip paralleled output power devices (Sankens, etc.)? If so, I would like to try it out.
Hoping that these questions could bring about a respectful, objective and constructive discussion.
Best,
Per
Get some Zen in and chill to the positive Chi from the east
tvrgeek (is that a reference to the great the sportscar I think it is?) did put in an effort to explain things which is much appreciated.
I am not familiar with the term Transient Miller Compensation (TM) - is that akin to the two-pole Miller I tried 5 years ago in post #105 on this thread? Or is it what Douglas Self calls output-inclusive Miller compensation?
Which I had to reverse to a standard feedback Miller after getting oscillations. At the time I concluded that it was due to the hand taped RA-820AX layout and component tolerances. No Spice simulation could have solved this, as tvr also rightly points out.
Then, there are different types of Dominant Pole Compensations (DP). Stan Curtis who designed most, if not all the Rotels always used a DP shunt topology, loading the VAS transistor collector with 33k in parallel with 330pF to ground. (Plenty schematics showing this at the start of this thread)
This heavy handling of the VAS is what produces the round/mellow "Rotel sound". It definitely stabilizes the amp, reducing after-sales service costs - but actually many people just loved it - some still swear by this sound.
I didn't mind it until Chris (post #119) tempted me to go for some 'fun and giggles' - which I did. And I have never looked back, nor has e.g. Paulus in his RB-991 upgrades (see post #394).
dreamth on the other hand suggested (at least I think he did) that doubling the drivers would be great. Yes, that will help the output stage drive, but wouldn't it also double the (non-linear) capacitive and current load on the previous VAS stage? What can/should be done about that - except for adding an extra driver stage? Also, to what level should these paralleled drivers be matched to ensure that one is not doing all the work while its partner is just freewheeling? Does anyone know if there are any integrated dual matched chip driver transistors - as there are internally chip paralleled output power devices (Sankens, etc.)? If so, I would like to try it out.
Hoping that these questions could bring about a respectful, objective and constructive discussion.
Best,
Per
"dreamth on the other hand suggested (at least I think he did) that doubling the drivers would be great. Yes, that will help the output stage drive, but wouldn't it also double the (non-linear) capacitive and current load on the previous VAS stage?"
You should look again at those posts and maybe download Kenwood KA-1000 and Kenwood L-A1 service manuals .You don't need additional explanations other than carefully studying those schematics.https://www.diyaudio.com/forums/solid-state/275171-improve-rotel-amp-thd-20db-46.html#post6314717
You should look again at those posts and maybe download Kenwood KA-1000 and Kenwood L-A1 service manuals .You don't need additional explanations other than carefully studying those schematics.https://www.diyaudio.com/forums/solid-state/275171-improve-rotel-amp-thd-20db-46.html#post6314717
First of all Chi or QI isn't about anything positive! Is about balance between positive and negative.Oh no, please don't start winding each other up again.........
Get some Zen in and chill to the positive Chi from the east
Hoping that these questions could bring about a respectful, objective and constructive discussion.
Best,
Per
Second...dominant pole compensation isn't about any filtering of any sort and for that i'll show some graphs as an image is better than a thousand words the chinese said.
Attachments
Last edited:
Oh no, please don't start winding each other up again.........
Get some Zen in and chill to the positive Chi from the east
tvrgeek (is that a reference to the great the sportscar I think it is?) did put in an effort to explain things which is much appreciated.
I am not familiar with the term Transient Miller Compensation (TM) - is that akin to the two-pole Miller I tried 5 years ago in post #105 on this thread? Or is it what Douglas Self calls output-inclusive Miller compensation?
Which I had to reverse to a standard feedback Miller after getting oscillations. At the time I concluded that it was due to the hand taped RA-820AX layout and component tolerances. No Spice simulation could have solved this, as tvr also rightly points out.
Then, there are different types of Dominant Pole Compensations (DP). Stan Curtis who designed most, if not all the Rotels always used a DP shunt topology, loading the VAS transistor collector with 33k in parallel with 330pF to ground. (Plenty schematics showing this at the start of this thread)
This heavy handling of the VAS is what produces the round/mellow "Rotel sound". It definitely stabilizes the amp, reducing after-sales service costs - but actually many people just loved it - some still swear by this sound.
I didn't mind it until Chris (post #119) tempted me to go for some 'fun and giggles' - which I did. And I have never looked back, nor has e.g. Paulus in his RB-991 upgrades (see post #394).
dreamth on the other hand suggested (at least I think he did) that doubling the drivers would be great. Yes, that will help the output stage drive, but wouldn't it also double the (non-linear) capacitive and current load on the previous VAS stage? What can/should be done about that - except for adding an extra driver stage? Also, to what level should these paralleled drivers be matched to ensure that one is not doing all the work while its partner is just freewheeling? Does anyone know if there are any integrated dual matched chip driver transistors - as there are internally chip paralleled output power devices (Sankens, etc.)? If so, I would like to try it out.
Hoping that these questions could bring about a respectful, objective and constructive discussion.
Best,
Per
Yes, I had a 74 TVR 2500M, (I converted to V8, C4, Posi, AC etc) And many more. Currently a B and a Stag.
It is the shunt on the VAS I was referring to. Thank you for his name. (Stan) I had it once and would like to give him credit as his choice was appropriate as I described where other designers chose other topology for their intended market. Erno used the same method in the Haflers, similar in the B&K, but being MOSFET, modification was a lot riskier. I do not know what was used in the "Trans Nova" but I remember them to be more transparent. Just loud transformers.
For sure, if you don't get it right for the full margins, you find all about ultrasonic oscillators. Briefly.
TM is how it was described to me. That was quite a few years ago to remember the details. There were several variations on Miller. I sold all the textbooks on the subject before I moved so I am not going back to research it. I built an amp I am happy with. By the time it fails, I'll probably be so old I can just go back to a Sony table radio.
Can’t recommend that, jvr. I’ve got a Sony table radio, and nothing but bad news comes out of it
And I never could afford a TVR, so I bought an old Datsun 260Z. Boy, was that car fast! I mean, from fully restored to rustbucket in less than a year!
Anyway, back on topic, I decided to revisit the output-inclusive Miller topology to see if it really is superior. This time on a RA-931 which is so much easier to work on than the RA-820AX.
Please note that I had already modified the standard RA-931 VAS by replacing it with a ”EF-VAS-EF” SMT module (see post #318) and removing the Rotel’s DP shunt components as part of upgrading it to ”AngelP Signature” (APS) level.
The VAS3 module now has an internal 27pF NPO feedback Miller capacitor from the VAS output to the first EF base. And with only that compensation the amp is perfectly stable – at least in my setups. However, as there is no way of predicting what speakers or load someone may connect, I now always add a 68pF PS around the entire module, quite easily done because Rotel already had prepared a pcb placement spot for such a Miller capacitor. See pic.1
I have measured THD at 1, 5 and 10kHz, both before and after adding the 68pF across the module. No measurable or audible difference.
Ok, then after a bit of drilling, track cutting and soldering, I managed to replace the 68pF with two 150pF PS in series with a 1k resistor from their midst to the negative feedback track from output. See pic.2
And measured before and after. Comparison results in pic.3 Things to note are:
Because my original goal was to make the APS upgrade available as a kit with fitting instructions so that people with only basic soldering skills would be able to do it themselves. Real diyAudio if you like.
Alas, I haven't got to that stage yet.
I managed to secure the web address below, but has still to find the time (and skills) to set up a working webpage. Having too much fun I guess, but moving forward at a steady pace....
Best,
Per
And I never could afford a TVR, so I bought an old Datsun 260Z. Boy, was that car fast! I mean, from fully restored to rustbucket in less than a year!
Anyway, back on topic, I decided to revisit the output-inclusive Miller topology to see if it really is superior. This time on a RA-931 which is so much easier to work on than the RA-820AX.
Please note that I had already modified the standard RA-931 VAS by replacing it with a ”EF-VAS-EF” SMT module (see post #318) and removing the Rotel’s DP shunt components as part of upgrading it to ”AngelP Signature” (APS) level.
The VAS3 module now has an internal 27pF NPO feedback Miller capacitor from the VAS output to the first EF base. And with only that compensation the amp is perfectly stable – at least in my setups. However, as there is no way of predicting what speakers or load someone may connect, I now always add a 68pF PS around the entire module, quite easily done because Rotel already had prepared a pcb placement spot for such a Miller capacitor. See pic.1
I have measured THD at 1, 5 and 10kHz, both before and after adding the 68pF across the module. No measurable or audible difference.
Ok, then after a bit of drilling, track cutting and soldering, I managed to replace the 68pF with two 150pF PS in series with a 1k resistor from their midst to the negative feedback track from output. See pic.2
And measured before and after. Comparison results in pic.3 Things to note are:
- The APS Rotel 1kHz distortion is already a very respectable -100dB (now an almost 30dB improvement!) and most residual is 2nd harmonic. Odd harmonics have vanished into the noise floor.
- Distortion figures do rise with increasing frequencies, but I have not found much difference between the two Miller topologies. No oscillations whatsoever.
- Yes, the "output-inclusive" VAS Miller typology is better overall, but at best only by a few dB.
Because my original goal was to make the APS upgrade available as a kit with fitting instructions so that people with only basic soldering skills would be able to do it themselves. Real diyAudio if you like.
Alas, I haven't got to that stage yet.
I managed to secure the web address below, but has still to find the time (and skills) to set up a working webpage. Having too much fun I guess, but moving forward at a steady pace....
Best,
Per
Attachments
Last edited:
With the risk of being off topic on offence once again, my daughter just copied me on this outstanding talk by Mr Bean. In full: Rowan Atkinson on free speech - YouTube
So, you don't have to be 98 to speak wisely....
I can understand the sentiment when we read or see the most ridiculous, e.g the "gay horse" or trivial offences being touted as "the new normal" in the media, but that's not the reality really is it? Do we really need to allow 1000s of offensive idiots to be allowed to devastate our society? Do we shut up the offended and weaker just for "free speech" which is wielded by the privileged and offensive. Shown by calling them "snowflakes" etc. if they dare to complain.
Free speech is an earned privilege not an automatic right, I'm afraid that we as a society have not earned that right yet. In fact sad to say we seem to be going backwards at an alarming pace.
Rowan Atkinson has just gone down in my estimation. The original sketch which I found hugely funny, satirical, apt and timely (then and now) was about racism not just trivial offences
Constable Savage: Racist Police (Not The Nine O'Clock News) - YouTube
One of the things we should all hold as valuable is the right to exist without any bigoted threat or offence, more than the right to be offensive.
But let's not deviate from a great thread about Rotel amps.
Free speech is an earned privilege not an automatic right, I'm afraid that we as a society have not earned that right yet. In fact sad to say we seem to be going backwards at an alarming pace.
Rowan Atkinson has just gone down in my estimation. The original sketch which I found hugely funny, satirical, apt and timely (then and now) was about racism not just trivial offences
Constable Savage: Racist Police (Not The Nine O'Clock News) - YouTube
One of the things we should all hold as valuable is the right to exist without any bigoted threat or offence, more than the right to be offensive.
But let's not deviate from a great thread about Rotel amps.
Yes, I am sorry for the deviation off topic.
Rotels have often frustrated and offended me - but I still love them.
Not many comments on the output-inclusive Miller issue. Maybe some of the LTSpice experts here on the blog could contribute by giving a simulation a try? Cabirio did some magnificent work (around post #201) and even generously provided his Spice file for us. So a lot of the footwork has been done already.
I'll be on to a Miller listening test today. (Not Glen Miller)
Per
No you are not privileged just offensive.
WTF are you threatening to shoot me?!!!!
What a nasty troll you keep proving yourself to be
I hope everyone knows now what sort of offensive person you are and treats you accordingly
offtopic again...
You insulted Rowan Atkinson's IQ...but he has something for you:
Hi Fi Shop Sketch - complete version - YouTube
Per
After the uncalled for bit of nastiness from the troll
I have always felt that the Rotel 991 was for some inexplicable reason lacking some magic which your changes along with the work of Paulus will unlock. My electronic knowledge is not even enough beyond changing components as instructed or for different types e.g. film caps resistor types etc. Many many years ago I had access to lab equipment but sadly no more I built my amp based on the graham nalty virtuoso in ETI and modified some cdp and sacd players, built a battery powered dac etc.
But I have done little original circuit design and so I am grateful for people like you who are willing to put there own hard work on here with clear explanations.
Thanks
Anjum
After the uncalled for bit of nastiness from the troll
I have always felt that the Rotel 991 was for some inexplicable reason lacking some magic which your changes along with the work of Paulus will unlock. My electronic knowledge is not even enough beyond changing components as instructed or for different types e.g. film caps resistor types etc. Many many years ago I had access to lab equipment but sadly no more I built my amp based on the graham nalty virtuoso in ETI and modified some cdp and sacd players, built a battery powered dac etc.
But I have done little original circuit design and so I am grateful for people like you who are willing to put there own hard work on here with clear explanations.
Thanks
Anjum
Here we go again.
I am tempted to just put on headphones and listen to Glen Miller's greatest hits for the rest of the day.
Or, I could copy you guys on the original design drawings for the Triple Energy Protective Shield used in the Battle of Naboo.
I use mine all the time, lowers the blood pressure far better than any medicine.
Per
I did not insult rowan atkinsons IQ (even Einstein made mistakes), just yours.as you have aptly demonstrated yet again by your childish response
So why do you feel that you are justified in your homicidal feelings and feel it's perfectly ok to threaten to shoot me or others like me???? Isn't that a threat or is that what you think is your god given right?
Please do not derail this thread any more than you have all along as some people.want to read and learn not listen to your ego.
You've started this derail mister whatever your name might be! I know people like you! Devious, passive agressive and f...annoying!
Do you have anything technical to say? If not , then just shut up!
Here's not your favourite LGBTQ hub , it's an electronics forum.
Last edited: