TGM Amplifier ?

[Lumba Ogir]

MJL21193,
pretty nice design, but how do you explain the function of U21/22/27 and the position of the CCS formed by U23/26?

[Bigun]

I looked at the data sheet for this bridge - no mention of soft-recovery. I may be overestimating the benefit of this feature and have zero experience in either case, but since I now have 25 of these diodes there's no turning back!

Yes, the LTP is certainly running lean. It's not that I'm excessively stubborn, but I see this as an adjustable item so my schematic doesn't imply that the final design will be that way. I still associate some of the 'black magic' of the final sonics to the way the LTP is set up.

I was hoping you wouldn't pick on C1, the pcb is squashed for space around there ! - for the HT application the DIY speakers will be set up with a -3dB at 80Hz - 100Hz will C1 get me down to 100Hz ?? somewhere I read that you need higher values if using electrolytic input caps (which I'm not) ??

Yes, most definitely plan to heatsink the VAS. I have a few options based on some heatsinks I've ripped off discarded circuit boards people threw out recently - worse case I'll screw on a simple piece of Al sheet and crinkle the edges.

I will have to check, but R3/30 partly helped, at the tim, with the pcb layout. As a fall back, I have some zero ohm resistors for camouflage !

R25/26 do need fixing don't they. Good thing I bought a 'selection pack' of resistors so that I can make last minute changes.

(I hope I remember to print out the PCB pattern in reverse)

And thanks for helping out, it makes all the difference.

[Bigun]

Regarding your design, I like it. Beats mine for total devices, like TGM 2 on steroids.

Why no charge suckout caps on the output devices, can C5 cover this need ?

You have no bootstrap, VAS is linarized with additional devices. Why do you pick this option ?

[mannycc]

...Linearity/accuracy is not the primary aim for me here but to sample the sound with some 2nd order harmonics from the imperfectly matched LTP...

I encountered the reason somewhere that circuit is made not balanced physically but is made balanced electronically meaning same amount of current should be passing on the EC of 2 input trannys for minimum distortion. Such case the value of resistors on input collector is important.

Regards,
mannycc

[MJL21193]

Originally posted by Lumba Ogir
MJL21193,
pretty nice design, but how do you explain the function of U21/22/27 and the position of the CCS formed by U23/26?

Hi Lumba,
U21,U22 are CCS for the VAS (U13,U12) - U24, U25 are current mirror.
U27 is bias supply for U21, U22 CCS.

U23, U26 set voltage for cascoded VAS.

Originally posted by Bigun
I looked at the data sheet for this bridge - no mention of soft-recovery. I may be overestimating the benefit of this feature and have zero experience in either case, but since I now have 25 of these diodes there's no turning back!

And thanks for helping out, it makes all the difference.

Hi Gareth,
I'm not a big believer in some of the more mystical aspects so the simple standard recovery bridge does me fine.

I have been where you are (still am, some would say ) and know the value of solid advise. Too often the new guy is left to take his lumps.

[MJL21193]

Originally posted by Bigun
Why no charge suckout caps on the output devices, can C5 cover this need ?

You have no bootstrap, VAS is linarized with additional devices. Why do you pick this option ?

C5 is the "suckout" cap.

The design grew out of another, covered here.

It's still in the planning stage and has been stalled for several months. I'll finish it eventually I expect.

[Bigun]

I'm not a big believer in some of the more mystical aspects

I shall be claiming superior sonics from these diodes and my special formula for the solder then

[Lumba Ogir]

MJL21193,
I see, however it`s not done correctly.

[MJL21193]

Originally posted by Lumba Ogir
MJL21193,
I see, however it`s not done correctly.

Hi Lumba,
Feel free to explain and elaborate here. Let's not gum up Gareth's thread with other issues.

Gareth,
Secret solder formula? Alchemy?

[Bigun]

Originally posted by mannycc


I encountered the reason somewhere that circuit is made not balanced physically but is made balanced electronically meaning same amount of current should be passing on the EC of 2 input trannys for minimum distortion. Such case the value of resistors on input collector is important.

Regards,
mannycc

That's correct, you have to look carefully at the current flow through each side of the LTP input stage rather than simply making each leg 'look' the same on the schematic. What I find helpful is the emitter degeneration resistors which in the SPICE simulation allow you to quickly compare the current flow balance by looking at the voltage drops across these two resistors.

I understand that these emitter degen resistors help with linearity, but not everyone likes them, fearing that they add noise.