Salas,
Does something like "hot rodding" Reflectors D is possible and makes sense?
I found that lowering R1 value and using bigger heatsinks Ref-D brings improvements in following case:
At first I used Soekris R2R with DCB1. It was great but then I thought why not to hot rod Ref-Ds powering Vref in Soekris. So used 1,5ohm (400mA) on Ref-D, while Vref needs only 50mA. I also put bigger heatsinks. They become a bit hot, but still you can touch them. Going lower with R1 value made sound more clear and heavy. Eventually I gave up on DCB1 and Soekris is connected directly to the amp.
Does something like "hot rodding" Reflectors D is possible and makes sense?
I found that lowering R1 value and using bigger heatsinks Ref-D brings improvements in following case:
At first I used Soekris R2R with DCB1. It was great but then I thought why not to hot rod Ref-Ds powering Vref in Soekris. So used 1,5ohm (400mA) on Ref-D, while Vref needs only 50mA. I also put bigger heatsinks. They become a bit hot, but still you can touch them. Going lower with R1 value made sound more clear and heavy. Eventually I gave up on DCB1 and Soekris is connected directly to the amp.
Hot Rodding makes some sense in shunt regulators. Increases the MOSFETS transconductance & makes the load current swings less % of the total bias current in M2 so its better suppressing rail distortion ability by the load. But it is always depending for effectiveness on the type of application (what kind of chips, what analogue stages, pcb layout, filters, stabilizations) and in total audio system synergy and preference as we saw JOIMONF using only small spare current.
Thank you Salas. With R1=5 Ohm M1 heatsink is barely warm🙂
Thanks for sharing.🙂 I will try if I have JFET with higher IDSS.
You can adjust a slightly too high Idss jFET to give a lower current by adding in a source resistor.
Hi,
Salas is it ok to put 1R2 in zobel network in Ref D instead of 1R? I do not expected any changes, just I have 1R2 under my hand. Wondering whether such value would not make any oscillation or limit bandwidth.
Salas is it ok to put 1R2 in zobel network in Ref D instead of 1R? I do not expected any changes, just I have 1R2 under my hand. Wondering whether such value would not make any oscillation or limit bandwidth.
I already fixed the Dac by remove the entire board plenty of jumpers, SMD, ICs, crystals, switch, dimm socket, etc. Can you believe?
Now as entire DAC I use only the 8 pin TDA1543, taking chance their hability to receive I2s signal (by 33ohm resistor). That's my Dac now, jajaa! Not only fixed, I got a more resolving DAC able again to fine tune the wonderful Refl-D supply!
Now as entire DAC I use only the 8 pin TDA1543, taking chance their hability to receive I2s signal (by 33ohm resistor). That's my Dac now, jajaa! Not only fixed, I got a more resolving DAC able again to fine tune the wonderful Refl-D supply!
In first, must apologize my first claim I reported lowering value the input cap CM to 1.000uF only (instead of 10.000uF.)
It was in my first essay back then, I compare every capacitor alone by themselves only prefered the detail and tone of small Nichicon KZ, instead of their bigger contenders also heard alone.
Maybe it was better not reporting such a quick round. I regret so much now after play again in deep, I found Nichicon KZ 1000 little bass response, but OK as bypass to 4700uF Nichicon KG.
Specificaly I found the best balance using paralel 4 caps: (main input cap: CM cap)
Nichicon KG gold aero, 4.700uF.
Nichicon KZ 1000uF.
Nichicon KZ 1000uF.
Paper in oil capacitor 4.7uF.
I tried one Elna Silmic II 2200uf instead of two KZ 1000uF, but not the same.
Neither increase more 1000uF KZ to reach closer of 10.000uF.
Going over 7.000uF strangle the high frequencies. AL LEAST to power my 50mA. Integrated IC. (not 140mA, as stated before, my fault also).
It was in my first essay back then, I compare every capacitor alone by themselves only prefered the detail and tone of small Nichicon KZ, instead of their bigger contenders also heard alone.
Maybe it was better not reporting such a quick round. I regret so much now after play again in deep, I found Nichicon KZ 1000 little bass response, but OK as bypass to 4700uF Nichicon KG.
Specificaly I found the best balance using paralel 4 caps: (main input cap: CM cap)
Nichicon KG gold aero, 4.700uF.
Nichicon KZ 1000uF.
Nichicon KZ 1000uF.
Paper in oil capacitor 4.7uF.
I tried one Elna Silmic II 2200uf instead of two KZ 1000uF, but not the same.
Neither increase more 1000uF KZ to reach closer of 10.000uF.
Going over 7.000uF strangle the high frequencies. AL LEAST to power my 50mA. Integrated IC. (not 140mA, as stated before, my fault also).
In the Mini & DC Flexy manual there is "For powering the Ref-D Mini a 9VAC/30VA transformer, four soft recovery diodes of choice, and
a 6800uF capacitor will suffice. Use higher uF if sharing a Flexy between a couple or more Minis. Especially if their CCS is set high." The four pole Mundorf pictured there is 6800uF.
a 6800uF capacitor will suffice. Use higher uF if sharing a Flexy between a couple or more Minis. Especially if their CCS is set high." The four pole Mundorf pictured there is 6800uF.
Hi Salas, is it okay to use 0.5 ohm for R1 if using large heatsinks for the mosfets (no heat problems)?
10V filtered DC input from separate LM338 supply. 5V DC output from Reflektor-D
10V filtered DC input from separate LM338 supply. 5V DC output from Reflektor-D
Even with enough sinking, for long term reliability reasons I would recommend Fairchild FQP3P20 for M1 which has higher ID current spec at 100C core temperature (-1.77A) where IRF9610 has (-1A) and you are going to produce 1.2A and 6W dissipation. The heat transfer rate from junction to case is two and a half times better in the Fairchild so it will remain cooler inside.
Thanks for your reply Salas! If using FQP3P20, and the load only uses over 1A for part of the time it's on (e.g. 500ma use for half of the time), are there other changes that should be made to Reflektor-D parts for best reliability?
Already checked the manual, thanks Salas also I read there on broup buy thread your advice for upgrade R1 resistor to Caddock MP, which makes sense as I regret so much having bought Mills, is waste money, it downgrade the quality of my homemade resistor.
I bring great satisfaction with C2, 1000uF Mundorf M-Lytic Ag +.
In comparison, brand new is similar to Nichicon KZ but enough half an hour of running, make it clear that mundorf is a beast apart.
In comparison, brand new is similar to Nichicon KZ but enough half an hour of running, make it clear that mundorf is a beast apart.
I was disturbed all night thinking about the reason why it sounded so different, I expected Mundorf AG + only a little better.
I listened again ... Nichicon KZ 1000uF has congestion in high frequencies, short opening, bass not articulated at all, no detail ... mmm .... Typical too much capacitance!
I put all the capacitor array Munforf KZ since 1000, down 470, 220, 100 and 47uF.
as always with the common mass and a good cooper alligator in the positive on plate, go listen!
It has been losing value and it began to improve until reaching 220uF, doubted myself ... sounds well enough or still able to improve? I put two parallel one hundred and yes better than 220uF, one might say that was ends up, but no, I tried valiantly 100 + 47. Serious and wonder with 'slam!' you know what I mean? only own the high quality, when the sound enters the stomach.
... from 1000 until 147uf. !!
Well, then I resolve by all, removing trash and get back to the crocodile but now only the 4.7 PIO and Teflon 0.01uF bypass.
After quick easy listen, these bypass will remain.
Bulky solution, but surprisingly has improved enough to reach close Mundorf M-Lytic AG+.
I listened again ... Nichicon KZ 1000uF has congestion in high frequencies, short opening, bass not articulated at all, no detail ... mmm .... Typical too much capacitance!
I put all the capacitor array Munforf KZ since 1000, down 470, 220, 100 and 47uF.
as always with the common mass and a good cooper alligator in the positive on plate, go listen!
It has been losing value and it began to improve until reaching 220uF, doubted myself ... sounds well enough or still able to improve? I put two parallel one hundred and yes better than 220uF, one might say that was ends up, but no, I tried valiantly 100 + 47. Serious and wonder with 'slam!' you know what I mean? only own the high quality, when the sound enters the stomach.
... from 1000 until 147uf. !!
Well, then I resolve by all, removing trash and get back to the crocodile but now only the 4.7 PIO and Teflon 0.01uF bypass.
After quick easy listen, these bypass will remain.
Bulky solution, but surprisingly has improved enough to reach close Mundorf M-Lytic AG+.
Last capacitor also critical in C1 place is 0,1uF.
But I found some congestion using modified russian telfon.
So SUBJECTIVELY, after several essay, I replace it for a 0,082uF + 4n7.
But I found some congestion using modified russian telfon.
So SUBJECTIVELY, after several essay, I replace it for a 0,082uF + 4n7.
I will try paste a picture how it looks the result, but only link to open next page available.
Power transformer is 15v.
I use a zener to output 8v./50mA. Still waiting to receive the LM329.
Temperature on radiators is 50ºC
Resistor R1 is 3R2 handwound isotan wire.
An externally hosted image should be here but it was not working when we last tested it.
Power transformer is 15v.
I use a zener to output 8v./50mA. Still waiting to receive the LM329.
Temperature on radiators is 50ºC
Resistor R1 is 3R2 handwound isotan wire.
Attachments
Hi Salas,
Can STP55NF06 be used as shunt MOSFET for a 3.3V Reflektor build, or only the L version can be used?
Can STP55NF06 be used as shunt MOSFET for a 3.3V Reflektor build, or only the L version can be used?