For both cases of R5=1K and 250R, Vo is around 10.1V when power up, drops to around 10.0V in 30 seconds, and then kept stable.
Is this 0.1V drop normal ?
Is this 0.1V drop normal ?
With R5=1K, Vo is lowest at 6.26V when trimmer is 0R, while Vf of the 3 LEDs are 4.97V. I didn't observe if it was stable though as first thing I did was to trim up the voltage.
I'll try again later, when I swap the 250R with 1K again.
I'll try again later, when I swap the 250R with 1K again.
Looks like it could be good for 5V too even with three LEDS positions. If one was shorted. What is it for? What will be the real load's current need?
Keeping highest possible CCS given the sinking ability is beneficial to its general characteristics BTW.
Keeping highest possible CCS given the sinking ability is beneficial to its general characteristics BTW.
I'm going to make 2 +10V Reflektor and stack them together to form a dual rail supply, it'll be used to power a buffer that I'm going to build. I don't really know the current need of this buffer as it isn't specified, but each channel comprises of mainly 4 x BC560C (plug some caps) so I believe the current need is pretty low.
That way much spare current will go to the output MOSFET and the output impedance will keep low which is a good thing. Stacking is good for avoiding ground loop also.
Hi Salas,
Sorry for my mistake. At that time I just took a glance at all your schematics for preparing parts, so I didn't make a decision for which version. Now I've built your design which is modded by quanghao. I used 3 sk30a, irf 9540 for positive and replaced my LM317+TL431 shunt power in my clone Naim Nac preamp.
I don't realize much improvements in sound and details when I play with medium volume gain, but for very very low vol. I've see much more details than before.
May I rite with my ears?
Sorry for my mistake. At that time I just took a glance at all your schematics for preparing parts, so I didn't make a decision for which version. Now I've built your design which is modded by quanghao. I used 3 sk30a, irf 9540 for positive and replaced my LM317+TL431 shunt power in my clone Naim Nac preamp.
I don't realize much improvements in sound and details when I play with medium volume gain, but for very very low vol. I've see much more details than before.
May I rite with my ears?
When your low level detail is more (resolution) it is usually also apparent in reverberation and sense of 3D when playing loud if the wave launch and lobes in the speakers are coherent enough.
My speakers now are Diy mini onken cabinet for dual concentric Sony driver (Fostex OEM) which I got from old CRT 29" Sony in 90's. For low volume gain it sound very good especially in mid range, suite for vocal, acoustic and instrument. But for loud volume it plays seem to be more screaming and becomes noisy.
After changing for super low noise power, I feel more comfortable when listening in medium gain. But speaker's ability is limited, so maybe I have to upgrade those.
Thanks Salas for your opinion !
After changing for super low noise power, I feel more comfortable when listening in medium gain. But speaker's ability is limited, so maybe I have to upgrade those.
Thanks Salas for your opinion !
I've done more testing and I take back what I said about output voltage being stable. I realize that the output voltage is actually swinging within a 0.01-0.02V range, I replaced the trimmer with a fixed resistor and the swing remains, and I can only maintain a completely stable output voltage (to the extent my cheap DMM can't detect anything) when using an all LED reference (i.e. shorting R6) or Zeners.
I thought the MOSFET maybe the cause so I replaced it with a MTP3055VL (others in this thread have used this MOSFET successfully before), but the swing persists. I checked the output with a scope meter and I don't find anything oscillation like.
What should I check next ?
If it takes a while to swing its the MOSFET's thermal changing the Iref reflected. And/or some BJTs in the mirror cycling. Larger VGS non TTL MOSFETs create more Iref so the percentage of change is smaller. How long it takes to swing that +/- 0.01V? For absolutely no move is using LEDS. Only very small R6 if any, just for hitting the target so it will not contribute enough. Oscillation, no. Problems would have been much stronger if so. You could use LM329 + 1 RED LED + short + short, the way you got your board made. That should be well near your 10V target in very good noise quality.
IMO I think noise is more important than volt stability of any psu, if it only swings around 0.05V tolerance. Just ignore it and relax with new toy.
cheers!
Need advice regarding Naim CD3.5 mods.
I'm doing basically all the stuff from Acoustica, but with slight changes - couple of Salas regs for TDA1305 chip, pins 1 and 28 (step 4 in the article). The question is should I keep 0.1uf Wimas or any other caps installed by Naim here on the pcb or stay with 220uf on the output of Salas v1 regs?
Also would it be a good idea to save some space for clock and other mods if I use same small Salas shunt as for lighspeed attenuator? Would it be a step over the standard LM317 solution installed by Naim? Alternatively here is another +5v reg from the same topic, how good is it if compared to initial Salas v1 reg?
One last question that doesn't refer to the regs, but if anyone would be so kind to assist I would very appreciate 🙂 It's coupling caps (step 6 in the article). Naim used 10uf tantalum caps between TDA chip and 3 output opams per channel. I assume Silmics or other electrolytics wouldn't be the best solution here, would they? I have the following options: Wima MKP, MKS, Myndorf M-Lytics 4.7uf, Solen 12uf MKPs (huge), what would be the best ones or should I order anything specific here?
Need advice regarding Naim CD3.5 mods.
I'm doing basically all the stuff from Acoustica, but with slight changes - couple of Salas regs for TDA1305 chip, pins 1 and 28 (step 4 in the article). The question is should I keep 0.1uf Wimas or any other caps installed by Naim here on the pcb or stay with 220uf on the output of Salas v1 regs?
Also would it be a good idea to save some space for clock and other mods if I use same small Salas shunt as for lighspeed attenuator? Would it be a step over the standard LM317 solution installed by Naim? Alternatively here is another +5v reg from the same topic, how good is it if compared to initial Salas v1 reg?
One last question that doesn't refer to the regs, but if anyone would be so kind to assist I would very appreciate 🙂 It's coupling caps (step 6 in the article). Naim used 10uf tantalum caps between TDA chip and 3 output opams per channel. I assume Silmics or other electrolytics wouldn't be the best solution here, would they? I have the following options: Wima MKP, MKS, Myndorf M-Lytics 4.7uf, Solen 12uf MKPs (huge), what would be the best ones or should I order anything specific here?
Thanks for showing me that part. I needed the TO-220 60W power though and have already made the order.
Sounds like some advice that only experienced Naim modders can reply. Maybe posting that again in the builds and fairy tales thread?
thanks! Already tested with couple of Salas shunts connected to the output! Wowww, that was an improvement!
It took a bit of time to order the parts and I tried the following:
- Use IRF640NPBF as M2;
- Replace all BC560C with 2N4403;
- Replace J1 (BF861A) with 2SK170;
- Replace J1 with a CCS diode E202;
- Put the whole Reflektor on a breadboard;
The time to swing varies, it's between few seconds to over a minute. I thought it's abnormal as the previous series regulator (using a single Zener as reference) I tried do not drift, but I'm starting to wonder whether it's actually normal to have such voltage drift for these shunts, as nokin has mentioned:
Attachments
- Status
- Not open for further replies.