Little GEM

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
This amp I called “Little GEM”, very simple circuit with TMC(with a twist) and very low THD. It uses one pair of the ThermalTrak ONSEMI output transistors.
Thermal compensation is divided in two part, output transistors with TT diodes and the drivers with Vbe multiplier mounted on separate small heat sink. I simulated thermal behavior and it looks promising, it will need find tuning during practical implementation.
I use similar circuit in one amp(normal output transistors) and it sounds exceptional good.
THD20k at 100W/8R is 0.000391% and at 4W/8R 0.000294%
THD1k at 100W/8R is 0.000047% and at 4W/8R 0.000029%
all simulated.
This amp it’s all you need for domestic listening, easy to build, standard easy obtainable components.
dado
 

Attachments

  • Little-GEM-sch.jpg
    Little-GEM-sch.jpg
    211.8 KB · Views: 2,050
  • Little-GEM.LAY].jpg
    Little-GEM.LAY].jpg
    179.3 KB · Views: 1,943
  • temp-sim.jpg
    temp-sim.jpg
    175.1 KB · Views: 1,763
well ...one pair of outputs @ 45+45 rails is a bit on the risky side ...what will happen in 4 Ohms loads ? especially if this is working without a VI limiter

then again doubling the mistake adding darlingtons will make the soar even worst ....

Lets say that the rest of the circuit looks far more promising but still you should recalculate everything with rails 35-40 volt

Again your simulator results might be a good point to start but to be useful you need to have a working prototype to understand how pcb , arrangement,implementation of the circuit and power supply will effect the zeroes for the worst ...

I am not trying to be negative but i am getting a bit fed up with all this simulation around me that has no actual use in real life ....Many forget to add what you correctly did that the impressive zeroes we see are simulator's not even result ...its just an estimation .

Kind regards
sakis
 
OK, we can keep power voltage at +-40 V and that is good for 4 ohm load. There are different wattage for ONSEMI TT transistors, so it could be scaled accordingly.
I seams that you don't like spice simulation. I've got very good exeriance with LTspice and practical result were close to the simulated one. I can't say it for THD as I never measured it on real amp, but the sound was the one to say it is good.
And one more thing regarding this amp, it is not only simulation(by the way simulation shows how good it could be is all role, as layout p.e. are followed) but if you read carefuly my previous post, I am listening a very similar amp.
Don't be fed up with spice simulation, with it you can design better amps.
dado
 
This one uses “normal” output transistors not ThermalTrak in OPS. It is the one I am listening now. The circuit diagram is very similar, technical specification the same as the one with TT output transistors, except how good and how fast it kips the bias current stable regarding temperature changes of the output transistors and in part the drivers.
dado
 

Attachments

  • Little-GEM-N-sch.jpg
    Little-GEM-N-sch.jpg
    208.9 KB · Views: 803
This amp surprises me every day, it do magic with the sound. For such simple circuit diagram it just should not did it, but it did. The bootstrapped VAS with beta enhancer and the TCM to do all the magic?? Moving C13(part of the TCM) from Q3 collector to the emitter halved THD20k. I tried that with my TT amp http://www.diyaudio.com/forums/solid-state/182554-thermaltrak-tmc-amp.html but the result was not so good. It could be for different reason, already very low distortion, JFET LTP, TRIPLE output stage, I am not sure, but here it did this magic.
Here is PCB layout for non TT version, I would like some comments on it as I tried to follow the role of minimal interference from the power supply. This amp simplicity is in pair with P3A. I have heard P3A some time ago and the sound was very good, but this one I like better, could be because it’s mine.
dado
 

Attachments

  • Little-GEM-N.LAY].jpg
    Little-GEM-N.LAY].jpg
    177.9 KB · Views: 758
Ah, that jogged my memory :) I have read that thread before... good stuff !

I've been noodling with a similar amp myself, so this has me interested. The main difference is I have an output triple (similar to Leach) and a regular current sink for the VAS rather than a bootstrap. I have been playing with TPC and TMC compensation, ironing out the stability is the sticky point.
 
NJL3281 are 180W or 200W devices depending on which misprint you read in ONsemi's datasheet.
Total Pdmax is either 360W or 400W.
Divide the lower by 6 (for running pretty hot) and the predicted maximum output power is 60W. Using the more optimistic of 400W divided by 5 (very large heatsink or domestic listening when the sink never gets hot) predicts 80W.

Those maximum power predictions are just about ideal for use with a 30Vac transformer giving ~+-45Vdc supplies, assuming a severe 8ohms speaker reactance.
 
post4 closed loop gain & phase, post12 open loop gain & phase.
They are quite different.
Could you show the .op for each and explain the plots we see?

Post12 is Loop Gain not Open Loop Gain and in the rectangle you can see Phase Margin, point when the Gain plot cross 0 dB, and Gain Margin the poit where the Phase reach -180 degree. Higher values are better and here, I think, they are quite good.
Post4 Close Loop Gain shows -3 dB bandwidth.
Tell me what you want with .op? First post shows(third plot) how the bias current depends of temperature( separate drivers temperature and output transistor temperature). This plot was generated with .op
dado
 
NJL3281 are 180W or 200W devices depending on which misprint you read in ONsemi's datasheet.
Total Pdmax is either 360W or 400W.
Divide the lower by 6 (for running pretty hot) and the predicted maximum output power is 60W. Using the more optimistic of 400W divided by 5 (very large heatsink or domestic listening when the sink never gets hot) predicts 80W.

Those maximum power predictions are just about ideal for use with a 30Vac transformer giving ~+-45Vdc supplies, assuming a severe 8ohms speaker reactance.

This is exactly how I think.
 
This is exactly how I think.
since we are agreed on this, can I suggest you test into a 4r0 load for a few seconds to see how well the output voltage holds up. (-0.6dBV is good, -1dB probably indicates that lower resistance is not worth testing).
If this looks particularly good, are you willing to try a few seconds into 3r0, and/or 2r66?
This is a good test of how well the amplifier delivers current.
Any longer than a few seconds becomes a test of the heatsink and I don't need that. I can model it well with the data the sink manufacturers give out.

Are you aware of the -20dB switch technique I use for testing at, or near maximum powers/dissipations?
 
since we are agreed on this, can I suggest you test into a 4r0 load for a few seconds to see how well the output voltage holds up. (-0.6dBV is good, -1dB probably indicates that lower resistance is not worth testing).
If this looks particularly good, are you willing to try a few seconds into 3r0, and/or 2r66?
This is a good test of how well the amplifier delivers current.
Any longer than a few seconds becomes a test of the heatsink and I don't need that. I can model it well with the data the sink manufacturers give out.

Are you aware of the -20dB switch technique I use for testing at, or near maximum powers/dissipations?

Working prototype has quite small heat sink, and I don't have lower load then 4 ohm.
I have to look back when I tested previous version(with standard TMC) to see if I did the 4 ohm measuremen. You can look into this thread too. http://www.diyaudio.com/forums/solid-state/186981-bootstrapsccs-t-tmc.html
I know nothing of the -20dB
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.