F5 Turbo is posted

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Kink in plots understood

It turns out that all of the plots I have shown with "kinks" in them at high power levels were underbiased. Ie. there was no current thru the diodes at the quiescent state. Dumb and dumber. I had neglected to look at the diode currents in addition to the total currents through the MOSFETs.

The rightmost plot in post #560 does have 30mA through the diodes, and .925A through the MOSFETs.

With actual tests (not SPICE) I have found that a bias of about 1.25A through the MOSFETs is needed to get the THD below .005%. The voltage across the 0R5 source resistors is about 0.525 volts.
 
I have one have finished, but I am in the middle of testing fets and going on vacation next week. Ilquam will have some of the matched pairs to test by weekend, so maybe we will have something to compare to simulation.

I am making live measurements of a balanced 1/2 Turbo. By 1/2 Turbo I mean it has one MOSFET per rail and one MUR1510 diode per FET. Here is an image of the board and is attachment to 1/2 of a HeatsinkUSA 10.08" x 7" heatsink:
 

Attachments

  • DSC_0341.jpg
    DSC_0341.jpg
    246.1 KB · Views: 639
  • DSC_0342.jpg
    DSC_0342.jpg
    244.1 KB · Views: 651
Very nice. How do the measured numbers compare to simulated ones?
Overall there is decent agreement between the simulation and the actual circuit, but I am not sure I have enough experience with this circuit to answer the question definitively. I have some doubts about the accuracy of the MOSFET and MUR diode models. There are some issues with the LTSpice results I haven't figured out. Perhaps there are LTSpice convergence parameters that need tweaks.

One thing that both the simulation and the actual circuit show is that reasonably high bias (> 1A) is needed to get THD below .005%. Both show that the Turbo diodes greatly reduce THD at high power levels versus the non-Turbo at the same bias.

I have not yet done any listening tests of the version with diodes. I have been listening for more than a month to a nearly identical balanced F5 pair without the diodes. It sounds very good.
 
Member
Joined 2012
Paid Member
One thing that both the simulation and the actual circuit show is that reasonably high bias (> 1A) is needed to get THD below .005%. Both show that the Turbo diodes greatly reduce THD at high power levels versus the non-Turbo at the same bias.

Since we are squeezed to limit bias and source resistance, has anyone considered 4 diodes in series-parallel? That way the people with enough heat sink can push the bias up higher and still keep the diodes from conducting at idle. There would be about 0.6 volts more drop at high power (depending on temp and current), but still have more power and less distortion than no diodes at high power. My main drawback to that is the extra cost of double the diodes and heat sinks.
 
No, the BA1 front end is voltage in and voltage out, output stage is push-pull.

For a current source amp, the output stage can not be symmetrical (top & bottom look the same)
F1 and F2 are current source amps, constant current source on top, bottom is transconductance stage (volt in, current out).
Current difference between top and bottom goes to the loudspeaker, so volt in and current out.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.