Do you have Cdom set up like in the schematic at the beginning of this thread? I would try moving Cdom to the VAS collector, that looks iffy to me (I think the benefit would be small anyways). Also, I think you should do away with two-pole compensation until you get it stable, then move up to more complex stability compensation.
Oscillation in the hundreds of KHz is strange What are you using as outputs? Those don't look like japanese devices... They may have horrendous Cob.
- keantoken
Oscillation in the hundreds of KHz is strange What are you using as outputs? Those don't look like japanese devices... They may have horrendous Cob.
- keantoken
Member
Joined 2009
Paid Member
I have only implemented the simple Cdom at this point. The output devices are Fairchild, in fact all the parts are from the same batch or lifted off the pcb of the TGM1. 125kHz seems low for this to be related to the CFP, it'll take a lot capacitance to fix it at this frequency - I'm going to check the pcb for the possibility that there's still some gotcha I've missed.
Member
Joined 2009
Paid Member
My intuition was correct, this is too low a frequency to be the output stage oscillating. The issue was rather disappointing, it was classical feedback from output to input due to the temporary positioning of wiring external to the amplifier. My input and output were too close and I can move the wires to turn on and then turn off the oscillations. So, onward and upward 
Member
Joined 2009
Paid Member
Found some >10MHz oscillations at the CFP driver, heating up the drivers and warming the Zobel. Fixed with a 68p on pre-driver (didn't try a smaller cap but likely can be reduced)
I've now listened to the amplifier. Most satisfying moment when you get the First Sound
Not sure how it is going to fare compared with TGM2. I removed the special mod from my TGM2 that generates H2 + H4 and did a quick comparison. They are comparable, but TGM2 may have the edge at the low end. More listening will be required.
I've now listened to the amplifier. Most satisfying moment when you get the First Sound
Not sure how it is going to fare compared with TGM2. I removed the special mod from my TGM2 that generates H2 + H4 and did a quick comparison. They are comparable, but TGM2 may have the edge at the low end. More listening will be required.
Member
Joined 2009
Paid Member
Initial listening tests are now done. The difficulty in listening to a new amp you've just got working is that you are heavily biassed towards wanting to like it.
I decided to restore the H2+H4 mod to TGM2 and listen to it compared with TGM3. The differences didn't seem much to start with. But I did find that TGM2 tends towards a punchier bass. However, for the rest of the music TGM3 sounds smoother. After awhile I found that I didn't want to swap back to TGM2, the sound of the new amp was quite a bit smoother, especially the upper mids.
Not trusting my judgement I asked my son who's a bit of a young piano player to listen to two amps and tell me which one he liked the best. He picked TGM3, although he didn't know which was which. So it looks like the Singleton + Triple is a keeper. That gives me 3 channels (out of 5 needed) completed and one bare pcb for TGM2 available should I want to use it.
So, where next, should there be a TGM4 for a 5th channel and what should it look like - perhaps a CFP singleton front end + Triple ???
I decided to restore the H2+H4 mod to TGM2 and listen to it compared with TGM3. The differences didn't seem much to start with. But I did find that TGM2 tends towards a punchier bass. However, for the rest of the music TGM3 sounds smoother. After awhile I found that I didn't want to swap back to TGM2, the sound of the new amp was quite a bit smoother, especially the upper mids.
Not trusting my judgement I asked my son who's a bit of a young piano player to listen to two amps and tell me which one he liked the best. He picked TGM3, although he didn't know which was which. So it looks like the Singleton + Triple is a keeper. That gives me 3 channels (out of 5 needed) completed and one bare pcb for TGM2 available should I want to use it.
So, where next, should there be a TGM4 for a 5th channel and what should it look like - perhaps a CFP singleton front end + Triple ???
Last edited:
Interesting, is there a possible correlation between 2nd harmonics and punchy bass?
I do know one thing; input and VAS stages that are conducive to 2nd harmonics tend to have very strong drive power in one direction, although not in the other. For instance, a singleton input will increase in linearity AND gain, and thus drive power when you increase the current. But when you decrease the current, the opposite happens and then it cuts off at zero. However looking at the LTP, either way you're screwed unless you simply have lots of OLG. I suspect this increase in drive power gives better bass linearity where it counts.
It is possible, with alternate topologies to the LTP and singleton input, to have this effect at positive as well negative excursions I believe. Maybe you could try a dual-singleton type input, as in the same concept as the dual LTP? This way you have tons of drive power right from the input stage in both directions. Might even need to protect the VAS.
Of course then the cherished 2nd harmonics begin to decline.
Here is the schematic to the TMG2:
http://www.diyaudio.com/forums/solid-state/145652-tgm2-amplifier.html
- keantoken
I do know one thing; input and VAS stages that are conducive to 2nd harmonics tend to have very strong drive power in one direction, although not in the other. For instance, a singleton input will increase in linearity AND gain, and thus drive power when you increase the current. But when you decrease the current, the opposite happens and then it cuts off at zero. However looking at the LTP, either way you're screwed unless you simply have lots of OLG. I suspect this increase in drive power gives better bass linearity where it counts.
It is possible, with alternate topologies to the LTP and singleton input, to have this effect at positive as well negative excursions I believe. Maybe you could try a dual-singleton type input, as in the same concept as the dual LTP? This way you have tons of drive power right from the input stage in both directions. Might even need to protect the VAS.
Of course then the cherished 2nd harmonics begin to decline.
Here is the schematic to the TMG2:
http://www.diyaudio.com/forums/solid-state/145652-tgm2-amplifier.html
- keantoken
Member
Joined 2009
Paid Member
I'll tidy up the end of the TGM2 thread by posting the final schematic
http://www.diyaudio.com/forums/solid-state/145652-tgm2-amplifier-13.html#post2202912
http://www.diyaudio.com/forums/solid-state/145652-tgm2-amplifier-13.html#post2202912
Member
Joined 2009
Paid Member
The method I used to inject H2 into TM1 was to feed some of the output to the top of the LTP and modulate the current. I don't know if this is a good idea or not but I've abandoned TGM1 as sounding inadequate. I believe TGM2 has punchier based because of the CFP in the LTP, it linearizes the larger excursions which tend to happen when there's a large bass signal. Injecting H2 means making the waveform asymmetric which means that the waveform moves more in one direction than in the other - Kean has described this perfectly. Whether it hurts bass I don't know, but H2 injection into TGM2 (done differently than with TGM1) still leaves TGM2 with the punchier bass of the 3 designs - although TGM3 is close enough to make it the overall winner so far.
Last edited:
Interesting design concepts in that design, try the single cfp and see how this works out. Have you seen the renardson amp, he uses feedworward feedback to virtually kill all THD especially anything above 2nds in his latest design, i ve built his simpler version just out of curiosity, its a very good amp, I just dont like the output cap configuration.
Member
Joined 2009
Paid Member
Member
Joined 2009
Paid Member
It looks very interesting all the same and a nice use of the CFP topology. It's a good suggestion, but I think I will save this for a different project as the TGM is not really set up to allow for Class A type of dissipation.
I have noticed a new book out, by Esa Meriläinen ("EMT" on this forum) which has inspired me. I haven't bought this book yet but it looks very interesting. Could TGM4 benefit from some of these ideas ?
I have noticed a new book out, by Esa Meriläinen ("EMT" on this forum) which has inspired me. I haven't bought this book yet but it looks very interesting. Could TGM4 benefit from some of these ideas ?
If you're asking, try putting a resistor in series with your speakers right now and see how it sounds. It sounded better to me, I have them in right now (it may have an affect on the amp). I was told current drive was best approximated by a series resistor of at least 10x the speaker impedance, but all I've got are 4 ohm resistors. Careful of the bass.
- keantoken
- keantoken
Member
Joined 2009
Paid Member
I tried a lowly 10R but it sucked the life out of it. The speakers are my own design, an Onken enclosure that is designed to flatten out the impedance peak in the bass region.
I got to listening to the amp again today. I threw a lot more material at it. I couldn't detect any fatigue that I might have experienced with the other designs and with my Bryston.
It sounds well behaved in the bass, not exagerated and not mushy but simply strong and smooth. The mids are more refined than my other amplifiers; voices and guitar strings sound as if they are right in front of me (I'm listening mono). The high's are not as clinical as the Bryston, but everything is clean - there is at least one song with some high frequency percussion - probably a metal triangle - and it sounds fantastic.
Right now it would be too soon to start on another TGM channel design because it's not obvious to me what I should be trying to improve any more. I need to wait until the initial impressions of this design are replaced by more critical listening. But TGM4 may well be one of Keantoken's Rush Cascodes bolted on to the TGM3 output structure.
It maybe that I will not be able to advance until I move to a higher power amplifier where performance limitations are more obvious.
So I will leave this design as-is, it is the first design of mine that I can now recommend to someone else to build.
A photo of completed module.... note the home made teflon capacitor for the phase lead compensation (twisted brown wires)
I got to listening to the amp again today. I threw a lot more material at it. I couldn't detect any fatigue that I might have experienced with the other designs and with my Bryston.
It sounds well behaved in the bass, not exagerated and not mushy but simply strong and smooth. The mids are more refined than my other amplifiers; voices and guitar strings sound as if they are right in front of me (I'm listening mono). The high's are not as clinical as the Bryston, but everything is clean - there is at least one song with some high frequency percussion - probably a metal triangle - and it sounds fantastic.
Right now it would be too soon to start on another TGM channel design because it's not obvious to me what I should be trying to improve any more. I need to wait until the initial impressions of this design are replaced by more critical listening. But TGM4 may well be one of Keantoken's Rush Cascodes bolted on to the TGM3 output structure.
It maybe that I will not be able to advance until I move to a higher power amplifier where performance limitations are more obvious.
So I will leave this design as-is, it is the first design of mine that I can now recommend to someone else to build.
A photo of completed module.... note the home made teflon capacitor for the phase lead compensation (twisted brown wires)
Attachments
Last edited:
Member
Joined 2009
Paid Member
Here's the updated schematic showing the as-built version.
(there maybe some hidden benefits in terms of sonics from the use of Tantalum as a bypass for the feedback shunt capacitor cunningly hidden underneath the pcb - highly recommended if the sound I've heard is anything to go by)
(there maybe some hidden benefits in terms of sonics from the use of Tantalum as a bypass for the feedback shunt capacitor cunningly hidden underneath the pcb - highly recommended if the sound I've heard is anything to go by)
Attachments
Last edited:
The bias generator for the input transistor looks really confusing to me. You might want to redraw it.
Tantalum has some really strange things going on. There used to be a web page up showing scope tests of capacitors (paper/oil was best), tantalum had some weird spiky things going when stressed fully.
I like the finished amp, and by your description I'd probably like the sound too. Currently I've got this Sony Xplod amp, it may in fact be the cause of the change in sound with series resistance.
Onkens, AFAIK, were engineered to be used with voltage drive; a series resistor would be improper. My speakers don't have any sort of crossover or special enclosure going; just fullrangers on a 6"X3" open baffle. So I guess it really depends on how much of the design has already been done for you.
Are you going to put PCB files somewhere?
- keantoken
Tantalum has some really strange things going on. There used to be a web page up showing scope tests of capacitors (paper/oil was best), tantalum had some weird spiky things going when stressed fully.
I like the finished amp, and by your description I'd probably like the sound too. Currently I've got this Sony Xplod amp, it may in fact be the cause of the change in sound with series resistance.
Onkens, AFAIK, were engineered to be used with voltage drive; a series resistor would be improper. My speakers don't have any sort of crossover or special enclosure going; just fullrangers on a 6"X3" open baffle. So I guess it really depends on how much of the design has already been done for you.
Are you going to put PCB files somewhere?
- keantoken
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.