Member
Joined 2009
Paid Member
too early to say about many things. I installed 47pF base-collector caps on pre-drivers as per the design, I did't want to take any chances on first build. I've used this output stage before (TGM3) also with these caps and never had any issues with stability. On my next build I can try without. The sound is very balanced, bass sounds punchy and treble sounds detailed without fatigue. Well although I have now heard it, much work remains as I have to make proper power supply, finish chasis, build two more channels and do some measurements. I think I will need to install some heatsinking to the main output drivers. But that now seems like the downhill part of it as I know there will be a nice sound at the end of this 
Thanks everyone for your support !
Thanks everyone for your support !
Member
Joined 2009
Paid Member
Lots of cutting and drilling going on today - making progress with the Chasis. Installed the heatsink on a pair of DIY flanges. Also installed some L-bracket that I can mount the PSU pcb on when it's ready.
RCA inputs also assembled, 3 dc-connected inputs and 3 ac-connected inputs via some nice 2.2uF capacitors I had been saving up for his. It's a point-to-point wiring job on this one but it will give me the chance to dc or ac couple depending on the source.
RCA inputs also assembled, 3 dc-connected inputs and 3 ac-connected inputs via some nice 2.2uF capacitors I had been saving up for his. It's a point-to-point wiring job on this one but it will give me the chance to dc or ac couple depending on the source.
Attachments
Member
Joined 2009
Paid Member
The power supply is coming along now...
Basic C-L-C, each C includes a parallel 150nF film cap. Soft recovery rectifier diodes, separate rectifier for +ve and -ve rails. Rail inductors are heavy duty 11A 0.33mH mounted on the back of the board.
Lots of wires.
The unused pcb next to the power supply is for future use - most likely dc-protection circuit for speakers.
Basic C-L-C, each C includes a parallel 150nF film cap. Soft recovery rectifier diodes, separate rectifier for +ve and -ve rails. Rail inductors are heavy duty 11A 0.33mH mounted on the back of the board.
Lots of wires.
The unused pcb next to the power supply is for future use - most likely dc-protection circuit for speakers.
Attachments
Member
Joined 2009
Paid Member
Basic chasis is now 'together'. Protective earth installed. An Alps volume (using 3 of the 4 wafers) pot and a centre notched alps balance pot added in parallel for the centre channel - to allow the centre channel to be adjusted up/down relative to master volume.
I had some bother last night when I powered it up - smoke released from a resistor and destruction of collector fuse on -ve side of amp. Replaced resistors and fuse. I thought it was parasitic oscillations from the CFP driver, but today I measured a short from -ve rail to the heatsink. Not sure of the cause, I dismantled the amp from the heatskink and reassembled it after cleaning around the pcb mounting holes, removing a piece of swarf. Seemed to fix the problem.
Got the CRO out. Clean 1kHz square wave with (8 Ohm) and without a load. I wasn't brave enough to try the square wave + capacitor-load torture test yet - I'd like to hear it again before I blow it up.
A quick look at the overall gain is attached (measured at discrete points from 2Hz to 500kHz) - the input filter rolls off the signal before it triangulates - and that happens past 0.5MHz....
I had some bother last night when I powered it up - smoke released from a resistor and destruction of collector fuse on -ve side of amp. Replaced resistors and fuse. I thought it was parasitic oscillations from the CFP driver, but today I measured a short from -ve rail to the heatsink. Not sure of the cause, I dismantled the amp from the heatskink and reassembled it after cleaning around the pcb mounting holes, removing a piece of swarf. Seemed to fix the problem.
Got the CRO out. Clean 1kHz square wave with (8 Ohm) and without a load. I wasn't brave enough to try the square wave + capacitor-load torture test yet - I'd like to hear it again before I blow it up.
A quick look at the overall gain is attached (measured at discrete points from 2Hz to 500kHz) - the input filter rolls off the signal before it triangulates - and that happens past 0.5MHz....
Attachments
Last edited:
Member
Joined 2009
Paid Member
Member
Joined 2009
Paid Member
- thanks gentlemen
Member
Joined 2009
Paid Member
Hi Gareth, really a great job. I particularly like your surface mount heat sinks. The three PCBs actually fits your heat sink perfectly, was this planned or coincidental
Thank you!
Yes, the pcb size was planned to make use of the heatsink I had ... I like to recycle and this heatsink came out of an old Onkyo amp (the same amp also donated the bottom panel and cover to the chasis, plus the volume and centre channel volume pots); the mains fuse, IEC power connector and trafo's all came from two old Acopian power supplies, the psu pcb from my earlier TGM project and all the internal wiring from old computer power supplies
Member
Joined 2009
Paid Member
Thanks !
Well I'm not sure my measured gain curves are that accurate, but I've run the simulation with current parts values and I've attached a snapshot of the simulated plot alongside my measured curve. Not sure I understand your question on cascode - but to confirm: I am using the full Hawksford front end, as proposed by LazyCat, which incorporates cascodes to the input devices.
It looks as if my measured results are close to the simulation and the overall gain is spot-on (x22 = 27dB). I'd have to check back to my actual numbers - I only have access to the plot that I posted right now.
Attachments
Member
Joined 2009
Paid Member
Yes, Lazycat has a high power version with cascoded VAS. I didn't go that route although I was sorely tempted but you have to stop somewhere !
However, you will notice that I did use an unusual amount of emitter degeneration for the VAS, eating up several volts of headroom in doing so. This is something I noticed that Bryston do on their 2B amp, another symmetrical design. It reduces sensitivity to temperature / device variations.
However, you will notice that I did use an unusual amount of emitter degeneration for the VAS, eating up several volts of headroom in doing so. This is something I noticed that Bryston do on their 2B amp, another symmetrical design. It reduces sensitivity to temperature / device variations.
Member
Joined 2009
Paid Member
Well I have all 3 channels powered up lying on the bench to see how stable the bias is. I've set it up for around 60mV across the emitters of the output devices and after an initial warm up it changed over a period of half an hour, drifting up 2mV to 4mV; not ideal given that I want a negative temperature coefficient but not enough to suggest any kind of thermal runaway risk as far as I can see. The dc offset moves a bit from cold but I didn't see anything to worry about.
And now for something controversial. Do I want the amp to be fully symmetrical ? The original SSA thread talked about a dominant 2nd harmonic, but simulations say otherwise if one has reasonably well matched devices, something the all-BJT version approaches much more closely than is usually possible with FETs.
Would it be better to tune the behaviour slightly so that we know the amp is slightly asymmetric and to do it in a controlled manner so that it's consistent between channels. Why would I want to do this ? - nominally I prefer a Hiraga-like sequence of falling harmonics with 2nd slightly higher or equal to 3rd.
A little asymmetry in the value of the series feedback resistors is all that is needed to fix a controlled amount of consistent asymmetry in the waveform. Simulations indicate that it's a clean method of doing so, it doesn't exacerbate the higher order harmonics - a problem with the behaviour of TGM1.
Attached shows the harmonic profile at 1kHz, +/-1V input into resistive load. The plot on the left is the output for the symmetric amp, the plot on the right for the 'tuned' amp.
And now for something controversial. Do I want the amp to be fully symmetrical ? The original SSA thread talked about a dominant 2nd harmonic, but simulations say otherwise if one has reasonably well matched devices, something the all-BJT version approaches much more closely than is usually possible with FETs.
Would it be better to tune the behaviour slightly so that we know the amp is slightly asymmetric and to do it in a controlled manner so that it's consistent between channels. Why would I want to do this ? - nominally I prefer a Hiraga-like sequence of falling harmonics with 2nd slightly higher or equal to 3rd.
A little asymmetry in the value of the series feedback resistors is all that is needed to fix a controlled amount of consistent asymmetry in the waveform. Simulations indicate that it's a clean method of doing so, it doesn't exacerbate the higher order harmonics - a problem with the behaviour of TGM1.
Attached shows the harmonic profile at 1kHz, +/-1V input into resistive load. The plot on the left is the output for the symmetric amp, the plot on the right for the 'tuned' amp.
Attachments
Last edited:
Member
Joined 2009
Paid Member
Some background. I've been working with a friend on a tube amp. He's been doing all the work as he develops a pair of SE mono-blocks. The experimental set up I've been listening to over the past few weeks has evolved. Mostly I've been listening to the russian equivalent of a 2A3 and it's been a real pleasure. Nothing I've got in SS matches the sound. It's not a tubey sound, it's clean but alive.
Well with the clean electricity one gets late at night along with good FM reception I've been listening to the TGM5 employing my 'tuned' up harmonic profile. It's the closest I've yet heard from a SS amp to what I heard from the SET mono block.
Well I'll stop there, builders pride always emphasizes the good in the sound of a new amp but I'm impressed with what I'm hearing. The sound is alive, but clean from bottom through mids to highs.
There is absolutely no turn on noise, nothing, nada, it's really quiet. No turn off noises either.
Well with the clean electricity one gets late at night along with good FM reception I've been listening to the TGM5 employing my 'tuned' up harmonic profile. It's the closest I've yet heard from a SS amp to what I heard from the SET mono block.
Well I'll stop there, builders pride always emphasizes the good in the sound of a new amp but I'm impressed with what I'm hearing. The sound is alive, but clean from bottom through mids to highs.
There is absolutely no turn on noise, nothing, nada, it's really quiet. No turn off noises either.
I don't agree with such a method: increasing pair distortion to hide impair ones.
Better to work on the curents and poles bandwitchs in order to reduce higter ones:
www.esperado.fr - Le crescendo revisité
In the same spirit, i do not agree with a tube amp sound as the reference: Why not to try the contrary ;-)
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.