I used 4700uF Nichicon KG with Qspeed diodes, and on the shunt PCB 220uF Elna Cerafines. Been very impressed with the latest black-skinned versions of the caps. Guessing they've been subtly improved over the years. Actually prefer these to Black Gate standards now. Looks like I got lucky with the component synergy - really powerful, but natural reproduction. Hugely involving. Dare I say it, the DAC is sounding pretty damn close to well mastered vinyl now
You can't run an Aston Martin on dirty petrol, can you? Its the same with PSU juice. Must be clean. When shunt is high octane also.
This config should do theoretically. For 75Vin & 38Vout R9 should be 24K, R6 8.8K, C2 50V. For 75Vout R6 can't be a trimmer reliably, for 38V a 0.5W trimmer will not suffer initially but better not for long term. FAIRCHILD KSA1381/KSC3503 can be found in large stockists, prefer ESTU hfe grade. Its power inefficient to use shunt regs with such constant current over load demand allowances in those voltages, watch your thermal planning and M1,M2 sinking.
Attachments
Thanks Salas, at the moment there is Zener being used for both. Sure this will be much much better!
You can't run an Aston Martin on dirty petrol, can you? Its the same with PSU juice. Must be clean. When shunt is high octane also.
Absolutely! It's a turbo-charged beast from hell now
Thanks for getting me yet another step closer to the music
J
Hi all, this is my first post to this excellent thread.
I am going to construct +ve and -Ve Version 1 regulators for the input stage,VAS and pre-driver stage of a Krell clone I am building (still )and would like some advice please.
Would it advisable to build a + and - regulator for each channel eg 4 regulators or would one pair be OK for both channels?
What would be a ball park figure for biasing the CCS?
Would a 35-0-35 transformer be suitable, I want an output between 30V and 37V (Yet to be decided )
Thanks for your help.
Alan
I am going to construct +ve and -Ve Version 1 regulators for the input stage,VAS and pre-driver stage of a Krell clone I am building (still )and would like some advice please.
Would it advisable to build a + and - regulator for each channel eg 4 regulators or would one pair be OK for both channels?
What would be a ball park figure for biasing the CCS?
Would a 35-0-35 transformer be suitable, I want an output between 30V and 37V (Yet to be decided )
Thanks for your help.
Alan
Yes you can put several loads on 1 regulator. But I think it's not good idea to power the logic IC and the DAC especially the analog part from 1 regulator.
No problem in my experience to use multiple secondary trans to power several regulators.
Regards
Hi, Alan
One pair can go 90% the way, if the driver stages are heavily biased so they don't move individual current consumption too much.
150mA more than load consumption is a good point. You should decide config and know/measure the consumption of what it will feed.
The Tx sounds adequate for enough Vin to Vout distance.
Hi Salas, thanks for the reply.The drivers for the output stage will be supplied from the high current supply that also feeds the output transistors. The PCB's I have for the Krell look easy to split the tracks before the drivers, so the shunt regulator will only be feeding the input and VAS. Sorry if my wording is confusing.
Thanks
Alan
Thanks
Alan
Alright, use a small resistor in series like 0.1 to 1R when you will do the split of what you have to use the regs for, with a dvm and the drop across that resistor when you drive with a 1khz sine wave just before clip see with Ohm's law what those stages draw for current. Sum what is for each reg to feed, add 150mA for CCS setting figure.
Reflektor +3.3V with STP55NF06L
The gate is fully open at last.
I implemented STP55NF06L into Reflektor as Salas suggested, getting great improvement.
When I used with 2SK2586, the sound was harsh. Now there is nothing, but clear ,smooth and dynamic. Especially transparency and 3-D reality is remarkable.
I have adopted four wire configuration for the first time, and surprised that the effect is very big comparing to two wire.
Zobel value is not optimized for sound quality whereas there is no oscillation at all.
Thanks Salas and other Reflektor lovers.
Lets hope it will open the Mosfet gate less capacitive at least.
The gate is fully open at last.
I implemented STP55NF06L into Reflektor as Salas suggested, getting great improvement.
When I used with 2SK2586, the sound was harsh. Now there is nothing, but clear ,smooth and dynamic. Especially transparency and 3-D reality is remarkable.
I have adopted four wire configuration for the first time, and surprised that the effect is very big comparing to two wire.
Zobel value is not optimized for sound quality whereas there is no oscillation at all.
Thanks Salas and other Reflektor lovers.
Attachments
Zobel is 'not' optimized, or is 'now' optimized?
Salas, in the REFLEKTOR, just to confirm:
1. that the CCS current is (2N4403)Vbe/R1?
2. And this should be = load current plus shunt current
3. The dissipation on 9610 mosfet is (Vin-Vout)*(CCS current)
4. The dissipation on 640 mosfet is (Vout)*(CCS current-Load current)
Thus when adjusting the Reflektor for different optinal currents, the best way would be to change R1?
Your help much appreciated.
1. that the CCS current is (2N4403)Vbe/R1?
2. And this should be = load current plus shunt current
3. The dissipation on 9610 mosfet is (Vin-Vout)*(CCS current)
4. The dissipation on 640 mosfet is (Vout)*(CCS current-Load current)
Thus when adjusting the Reflektor for different optinal currents, the best way would be to change R1?
Your help much appreciated.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.