hagtech wrote:

It's difficult for me to guess exactly what the possible sonic effects of the ripples would be.  I see a tube operating at a fixed point, having a particular distortion spectrum.  Say -42dB 2nd harmonic.  Then, the ripple pushes it to a new operating point, giving -40dB 2nd harmonic.  So the treble signal is modulated where distortion fingerprint changes at a 120Hz rate.  Is that audible?  What does it sound like?  Maybe I am chasing ghosts. Similarly, we have the same thing with DHT, as the operating point of tube changes with the temperature of the cathode, heating and cooling at 60Hz.  Maybe this is so far down in the noise floor we cannot hear it yet.  There are other peels to the onion that must be removed first.

Jim, I do not think that it is the question. There is no argument that “no ripples at all” is better than ripples and no one argue that sonic effects of the ripples, no meter what it would be, is altogether negative. However, there is also a negative effect of the mechanism that minimizes ripples – the capacitance.  So, the quest is where the negative impact of ripples would be “competing” with the negative impact of the capacitance. Perhaps you took me too literary what I said about that “ripples do not meter”. So far in Super Milq I use huge a huge amount of capacitance with having absolutely microscopic amount of ripples – I would like to revise it for the HF one-stage amps. Not the least reason, and perhaps the main reason, is because, as I said before “larger capacitance makes HF harder”. Actually I even being to question if I even need a second cap in amplifier itself. In all my 2 chassis units, including the Super Milq I have the LC with a bleeder in PS and then RC in control unit. I usually try to keep the last cap as close to load providing the shortest DC path to ground. If I keep the last cap of small value on the PS side then I lose bass and Dynamics. Now, I wonder if I need to worry about it in context of HF one-stage amps. It might be that I will be go away with just one single LC on PS side with a small cap of very good quality (not metalized). I might be interesting if it turn out to be true… It always will be easy to add the second cap on the amp side to short the ground locally… BTW, since it will be effectively a high-pass filter and since I do not do LF it one-stage amps then might be a very small cap, perhaps the Teflon values…

I am trying to think about this power supply from a perspective that less is better.

Which caps exactly did you decide to try?  I don't recognize the big black jobbies.  Are they oilers?  Are they 200V?

Sticking with the Hammond chokes?

Best regards,
Paul S

Paul S wrote:

Which caps exactly did you decide to try?  I don't recognize the big black jobbies.  Are they oilers?  Are they 200V?

I did not “decide” but went with what I had in my hands. They are the Ansars from the picture few post above, Metallized Polypropylene, 100uF/450V. I use two in parallel.

Paul S wrote:

Sticking with the Hammond chokes?

Sticking with whatever I need. The Hammond chokes are the bias chokes 40H.  I like Hammond chokes BTW, they are better then many others. The 6C33C chokes are Toroidal 10H/400mA custom made by Toroidal Maryland. The first stage is Maryland custom as well.  The single-stages choke is 15H/150mA Lundahl. I was trying to use it before and hate it – never was able to make it to sound/measure good. It was literally a trash choke… until 2 weeks ago I paid attention that it has absolutely unintuitive marking of the coils and that I connected in past the cols … against each other. What a Moron!!! Anyhow I pulled it literally from a trash box and it works very well now.

hagtech wrote:

.  You have 30mA bleeder for each output?  Seems like a lot of draw.

Nope. It is not for each output. The Second stages do not use bleeders; the tubes suck enough current to warm up the chokes – 250mA-300mA against 10H. The bias chokes are 40H and each solder draw 20mA. I bleed another 8mA from each side. The first stage is 3 X 6E5P by 17mA each against 15H. I bleed 30mA extra.  The single-stage is 3 X 6E5P/6E6P by 30-35mA each against 15H. I bleed 8mA. All supplies with exception of bias are shorting on 50R resistor when the power is off. (I disconnect the tweeter that has a transformer with B+ aboard) The 12V is CRC full wave, completely lifted from chassis. I am not convinced that I will stay with the current single-stage configuration. My initial draft will be the second RC filtering stage in the amp side – 250R and 100uF the same Ansar cap to introduce a local path to ground. Well, will see how it works out…

I was playing with new power supply over this weekend, trying to optimize everything. First I got rid of bleeders at all B+ - it has enough critical current with them according to my measurements after the bride. All supplies after the first choke have no ripples but a perfect sinusoid. The AC after the second filtering stage in the all supplied except the single-stages is un-measurable. The single-stages B+ has after LCRC (200uf, 15H, 500R, 100uF) AC of 5mV. I will see if it is stay when I will be listening the thing – I might drive the capacitance lover.

Those Silicon Carbide diodes are very nice and I have an accidental stress testing. I made all PS to be shorted by 50R resistor as the amp turned off. I use 3 individual 5A relays that turned to be too flimsy for the task as the capacitance of the two stages supple is too large. So, one of the relays but welded and when I turned the amps next time the B+ of the first stage was shorted by 50R resistor. It is 75W aluminum resistor direct-mounted on the metal chassis (not visible on the picture above).  So, I turned the amps and went to shower. When I got back my room was full of smoke, I ran to the amp to shut it down, figuring that it should be all burned. The entire amp’s chassis was near 80-90C, I turned the amps off. The 50R resistor was the temperature of supernova. It was not burned through though and the diodes DID NOT BURN!!!!  The resistor after baking measured 33R. Well, I figured that 33R against 400V – so we are taking about the 12A of continue currant – not bad at all for 600V Schottky!!!

There is some open issuers as well. If to look between the bridge and the first choke the beginning of the waves look very strange. If you look at the picture then you will see that voltage go to zero as it shell (calibrated at first square), then there is a there is very small horizontal step and then, what the new wave states there is a very strange vertical half square rise. I wonder where it comes from. Is it some kind of anomalies of my toroidal chokes? Should I worry about it? The voltage after the choke gets converted into as perfect sinusoid anyhow… I do not know the answer yet.

Also, the 400V supply, driven into the toroidal choke show some choke ringing. I wonder if I need to fight with it. The choke overrides the ripples into sinewave anyhow and the second filtration stage kills any sights of AC anyhow. So, how bad the choke ringing and does it has any effect to sound? One of the advantages of use the very large last cap is that all “qualities” of PS become less relevant from sonic point of view. So, does the choke ringing matters?

Romy the Cat wrote:

Those who flow what is going on with my 6-Ch Milq know that for sniggle driver stage I this revision I went what for non-electrolytic capacitors. I use a mandatory for me input choke with 200uF on PS side and then 100R and 100uF on the amps side to ground. The caps I use British Ansar, metalized polypropylene 450V.

I have been listing this new MF channel for a few days. My typical way to access a MF channel is a long tradition for me: disconnect everything besides MF and Upperbass channels and listen string quartets. There are a lot of middle and late Beethoven’s quartets, Dvorak's, Bartok’s and Schubert's quartets is being played lately in my room.

What can I say about sound of the new MF channel? I hate it, I really do. At this point I do not know why it is. Perhaps the caps are garbage, perhaps the diodes are not good, perhaps the caps need time to break in, the OPT problems or perhaps I am experience of hundreds other problems that screw up Sound. I know that something is wrong. The MF channel sound now with the new amp very flat and not articulate, it has some very odd compression. It probably the worst sound I ever got out of S2 driver. Other might feel that it has the “field-coil softness” – not me. I call that softness as transience deficiency and the driver disability to go into the dancers regions.  It sounds almost like Altec 808 with symbiotic diaphragm driver by 2W amp - what would be more disgusting? It is also important that the “tone” of the driver also collapsed (I would like do not let me to get stated about the tone of my MF driver  now – it sounds like somebody is tapping atop of a turned upside down cooking pot!!!). I have also not idea why it is so and I have also not idea what is going on. I will keep to burn the Ansar caps for a next week or so then try to go back with my electrolytic to see if it will do a difference and return Sound where I must be.

Paul S wrote:

Romy, isn't this the first time you have tried the film caps for PS and the "ALL-NEW" "SUPER-SOFT" diode bridge?  I would certainly go back to known quantities before I did or un-did anything serious.  Not that I always follow the "1 Change At A Time" rule; I just usually wind up having to do it that way eventually... Meanwhile, I suppose you'd know for sure if the biggest problem was the effing electricity? Not trying to piss you off, honestly; my own bad electricity shows up with some new symptoms about every month.  I really should keep an electricity log.

It is not electricity for sure as I have left channel 6-way amp that I was running in identical configuration. The right and left MF (the same signal) sound not as two different amps but as two amps coming from two different planets. I need to investigate it further as something is obviously very wrong.

Well, it “might” be related to electricity as the left amp has single driver getting power from 15.000 cap. With 30mA load that channel works from a virtual DC buttery with this amount of capacitance and the cap never discharged. So, the PS with enormous capacitance usually more immuned to electricity problem that the PS what the power caps are constancy fully recharged. That was that a whole idea to use outsized caps in Milq in order to eliminate the contribution to sound of rectifiers, cables, chokes caps, transformer and the rest crap. Milq is less sensitive to electricity then my front gear, not immuned though…

I will look for the answers replacing the elements in the following order:

1)    S2 Driver
2)    Both filtering Caps
3)    Choke
4)    Rectifiers
5)    Transformer

It is kind of suck as with learning why I lost MF on the right channel I can not to move forward to finish the 6-Ch Milq project.

biggest problem was the effing electricity?

…as the similar channel on the left amps play perfectly fine and very differently. I did switch the S2 drivers and it had no result. The most funny is that the right channel has very plastic horn honk – where the hell THAT come from? I cannot stress how heavy that horn is 0 it is actually funny how it sounds. Any plucking of strings sounds like a toping of an empty shoebox. Tonally the channel plays virtually juts one note with a very minor deviation of volume or frequency. I begin even to think that I might check out the OPT as well. Who knows… 

I actually if I have time tonight will run a wire from well-sounding 200V from the left amp to the new right amp. It would be fun it after all I will learn that my high voltage silicon carbide diodes or the Ansar caps have that “Sound”. Anyhow, I would not put blame at this point to anything and I clearly do not know where this sonic misery came from…

Meanwhile the amp is running for 5 days already non-stopping that made the caps to have over 125 hours on them…  Let see what the new day will bring. It would be fun if another Audiophile myths turn out to be … bogus.

Well, the subject of capacitors in that PS should be resolved. We discussed it a lot with Dima and he feel that I am fool all my idea to try in this supply “minimum capacitance”. He actually does not believe in small capacitance, he has a vision how the last capacitor participate in signal path and according to him the last cap should be large. Dima feels that zillions of SET out there with 20uf or 60uF capacitance the plate sound as horrible as they sound because the imposer recharging pattern of the last cap (one of the reason) and therefore the PS are improperly decupled from tube plate. I do not know if it is correct and if it is play in my case. I never experience any problem with PS sound but I always used very large last cap. According to Dima a sufficiently-large last capacitor with using in input choke make the sound or transformer, rectifier and choke less relevant. It is possible that he is right as it was my very first PS where the last cap was less than 2000uF -3000uF. Perhaps now, with “minimum capacitance” the maternal of the caps begin to “sound”.

N-set, you said that switching of my diodes has a lot of room to be better; could you elaborate could/should I manage it?  Thanks.

a sufficiently-large last capacitor with using in input choke make the sound or transformer, rectifier and choke less relevant

… but there is no one to ask how mater it is…

hagtech wrote:

Seems to me a small film cap added to a large electrolytic solves most of the sonic issues people complain about re the large caps.

Hm, I do not know. I think if good caps used then the bypass might not be necessary. I am big fun of Nichikon capacitors, to me they are absolutely as good as capacitors should be. The Super Milq’s B+ for 6C33C uses 350V/15.000uF for each tube. A year or so ago I did bypass those big caps with 2uF polypropylene caps. I use the 590Cudes, the caps I know very well and that I trust to serve coupling duty. So, I bypassed the big Nichikon with Cubs…the 2uF was the right value… and I can NOT report that I heard any improvement or any serious difference. I decided to keep the Cubes in there but… only because I have spent efforts to put them in – there is no other auditable reasons.

As I said I can’t not judge if the excessive amount of last cap eliminates the sonic contributions of the elements before the caps. However this vision dose sounds reasonable to me. Evan if I change my 100uF Ansar to 3.000 Nichicon and get right Sound then would it be the amount of capacitate or the time of capacitate? I think I need to change the Ansar to 100uF Nichicon… no matter how much I hate those experiments….

I still wish that the problem would be with caps and it would be easier to replace or to add…