The tube depot and the tube store have 500V JJ's and a few other brands.
Can Capacitors
http://thetubestore.com/cancapacitors.html
Can Capacitors
http://thetubestore.com/cancapacitors.html
Checked those and still don't see excactly what we need. I'm afraid stacking may be the answer. I'll need to build on Vero board I guess.
TubeMack: You can use the LCR brand 200u/500V cap for C2 and the Pana TS-UP 120u/500 for C1 (both readily available) without any issues if you don't have the room to run caps in series. The ripple V will be slightly higher using a 200u in place of the 330u for C2.
Or, as Tubelab suggested, parallel a motor run with a 120u TS-UP Pana for C2.
Running the caps in series gives loads of voltage headroom compared to the above two solutions, though. I would go with Pana 400V TS-HA's in series. They have the lowest ESR of the Pana snap in series IIRC.
And as I mentioned earlier, JEA Capacitors in NY have plenty in stock [currently] with good qty discounts. This is a good eBay store. A large surplus operation and they sell only new stock. I have bought plenty from them.
Click here for current stock of 330uF/500V caps.
..Todd
Click here for current stock of 330uF/500V caps.
..Todd
Last edited:
TCI is probably a Taitron brand. Taitron is a California based electronic components ODM. An ODM designs and sells items that are made by someone else to their specs, and since Taitron have offices in China, one would expect that to be the home of their suppliers. (Methinks that Made in Usa tag may be the result of a legal labeling loophole, but that's just a guess.)
If you are familiar with Dayton brand speakers from PartsExpress, then you know an ODM. Same idea; Dayton designs the speakers then gets them made in China (or wherever) to their spec. It's a very common business model.
That doesn't tell us anything about quality, but I will be ordering these for the 330uF value just because it's the easiest/cheapest solution. I'll report back on any fireworks displays, but I expect them to work fine, and they are high-temp rated, which is good.
..Todd
If you are familiar with Dayton brand speakers from PartsExpress, then you know an ODM. Same idea; Dayton designs the speakers then gets them made in China (or wherever) to their spec. It's a very common business model.
That doesn't tell us anything about quality, but I will be ordering these for the 330uF value just because it's the easiest/cheapest solution. I'll report back on any fireworks displays, but I expect them to work fine, and they are high-temp rated, which is good.
..Todd
Last edited:
I never thought I'd see the day where the Americans look at worse supply then we do.
The whole 'having the money but still can't buy what I want' thing - quite soviet.
The whole 'having the money but still can't buy what I want' thing - quite soviet.
After reviewing my meager collection of HV caps and the Digi-Key web site, it has occurred to me that all of the Panasonic TS-UP caps that I have purchased in the past few months are still in stock at Digi-key.
I also re-ran the PSUD model with C2=220u, and the difference in ripple V is negligible; they both produce about 20mv ripple at the 465V tap, which is plenty low for a PP design.
So, for the single cap (non-series/non parallel) solution, going with C1=120u TS-UP 500V (Digikey P/N P7451-ND) and C2=C3=220u TS-UP 500V (Digi-Key P/N 7456-ND) would work fine.
This also saves space and $$.
Also, if you want to use a 500V TS-UP for C4, the smallest available value is 47u; this would also work fine.
In light of the above, I also ran the model with C4=1u and the ripple V on the 288 B+ is below the resolution of PSUD using both a 1u and 47u cap for C4. They are both flat as a pancake.......since there are 3 RC sections upstream filtering the daylights out of the B+......, so you could also use a 630V film 1u like a Solen or equivalent for C4.
The model attached is 120u-2.5H-220u-330R-220u-68K-1u.
Attachments
Hey Boywonder, I was playing with PSDU2. How do you derive the ripple current/voltage? I don't see it. If ones needs to calculate it via Ohm's law or a time constant, I don't know where to grab the values to figure it.
I was noticing that using C1 as low as 47uF is still acceptable too (before the output voltage is adversely affected), but I don't know what the downside(s) of doing that is--and there is always some downside. But I was thinking 100uF might save a couple more pennies with little or no downside.
..Todd
I was noticing that using C1 as low as 47uF is still acceptable too (before the output voltage is adversely affected), but I don't know what the downside(s) of doing that is--and there is always some downside. But I was thinking 100uF might save a couple more pennies with little or no downside.
..Todd
Last edited:
taj,
Just scroll over on the table. Make sure you set you window after the voltage has stabilized. The ripple is the third column iirc.
As far as C4, I will using a 5uF motor run. The ones I have are about 2" square and rated 440VAC. Allied should have them #591-0245 <$6.00 each
Just scroll over on the table. Make sure you set you window after the voltage has stabilized. The ripple is the third column iirc.
As far as C4, I will using a 5uF motor run. The ones I have are about 2" square and rated 440VAC. Allied should have them #591-0245 <$6.00 each
Hmm... Mine shows Min, Max, Diff, Mean and RMS columns. Maybe it's another Win7 problem. The help doesn't run, it crashes often and acts weird. I'll try it at work on my Windows XP computer to see if it's different.
..Todd
..Todd
I've got close to 20 Xicon 100u/350 caps (20X40mm). Most of them soldered into a circuit before, but still fine. Send a box with return postage and you got em.
Only cost will be a small contribution the the site. You decide the amount.
Sheldon
Only cost will be a small contribution the the site. You decide the amount.
Sheldon
Last edited:
For ripple voltage, just zoom in on the flat part of the curve; you will have to zoom in several times, then the y-axis scale gets fine gradients, and read off the PP ripple V. I'll have to play around to figure out ripple current. IIRC, Gingertube explained how to get ripple current on the baby huey thread but it presently escapes me.
In a CLC supply, with low C1 values below a certain value (usually around 10u), you head into the transition zone between a capacitor input supply (sqrt2*trans secondary V) and a choke input supply (approx .9* trans secondary V). So as you have noticed, you can dial in the B+ voltage between 1.41*transformer secondary voltage and .9*transformer secondary voltage by changing the value of C1 from less than 1u to around 10u.
Reducing the value of C1 reduces ripple current (tube rectifiers require small C1 values, SS rectifiers don't) so reducing C1 to 47u or 68u is an improvement in this design. I can't remember how we got to C1=120u, but my early PSUDs had C1=47u. We just want to stay away from values that start causing B+ to drop.
IIRC, some folks here have mentioned that using small C1 values to dial in the B+ voltage in the choke-to-cap input zone mentioned above makes voltage regulation worse than higher values of C1.
I just called Edcor and asked if they have the primary R, secondary R, or the offload voltage for the XPWR002. Sadly, they did not have any of the above.
I am a little worried about some of the arguments regarding the power supply capacitors. It is not only about ripple, but about a stable non-oscillatory voltage stabilisation under programme conditions.
Music consists of sudden peaks in a normally lower load situation. I am not familiar with your testing programs, but put a real variable square wave current load over say C3-C4 (f = 10Hz), and watch the resultant signal-ripple on the h.t.? One must remember that e.g. C6 doesn't additionally smooth ripple coming at it from upstream; it is supposed to keep the effect of 'surges' of audio off the EF86. A 32µF cap there has a. l.f. time constant of about 10 seconds, which means that 1 sec. variations would already start affecting the EF86 'isolation' from full signal effects on the power supply (we are working with class AB1 here). Not knowing the power transformer secondary impedance makes it impossible to calculate, but I would personally feel uncomfortable with anything below 32µF there, for what it is worth.
If I read correctly 5µF has been suggested for C4. The time constant for that with 220 ohm (R1) lies at 144 Hz! - it might as well be left out, except perhaps for some h.f. isolation. In fact, I am not sure that C4-R1 serve any purpose. The practical test for this is having a finished amplifier, and feed it with tone bursts of close to maximum amplitude and check with a 'scope on the various h.t. points [and stand ready to switch off in case any tendency toward l.f. oscillation (motor-boating) sets in].
The above arguments for C1 were sound. But again a high value could minimise hi-load/lo-load variations. Again a gut-feeling suggests >47 µF. I would not think that cost would be a factor. (The power transformer secondary winding impedance of some 10s of ohms would limit too high a peak charging current.)
These are musings but from some experience - as said before, if someone knowledgeable could build the first model and test. My main plea is that l.f. stability is more important here than hum. When power supply stability conditions have been met, any ripple would be below being bothersome. (I do not have your programs, but I usually slap a basic circuit on to Spice, and start modulating the load with signal bursts to see the effect. Then off to the real thing.)
Music consists of sudden peaks in a normally lower load situation. I am not familiar with your testing programs, but put a real variable square wave current load over say C3-C4 (f = 10Hz), and watch the resultant signal-ripple on the h.t.? One must remember that e.g. C6 doesn't additionally smooth ripple coming at it from upstream; it is supposed to keep the effect of 'surges' of audio off the EF86. A 32µF cap there has a. l.f. time constant of about 10 seconds, which means that 1 sec. variations would already start affecting the EF86 'isolation' from full signal effects on the power supply (we are working with class AB1 here). Not knowing the power transformer secondary impedance makes it impossible to calculate, but I would personally feel uncomfortable with anything below 32µF there, for what it is worth.
If I read correctly 5µF has been suggested for C4. The time constant for that with 220 ohm (R1) lies at 144 Hz! - it might as well be left out, except perhaps for some h.f. isolation. In fact, I am not sure that C4-R1 serve any purpose. The practical test for this is having a finished amplifier, and feed it with tone bursts of close to maximum amplitude and check with a 'scope on the various h.t. points [and stand ready to switch off in case any tendency toward l.f. oscillation (motor-boating) sets in].
The above arguments for C1 were sound. But again a high value could minimise hi-load/lo-load variations. Again a gut-feeling suggests >47 µF. I would not think that cost would be a factor. (The power transformer secondary winding impedance of some 10s of ohms would limit too high a peak charging current.)
These are musings but from some experience - as said before, if someone knowledgeable could build the first model and test. My main plea is that l.f. stability is more important here than hum. When power supply stability conditions have been met, any ripple would be below being bothersome. (I do not have your programs, but I usually slap a basic circuit on to Spice, and start modulating the load with signal bursts to see the effect. Then off to the real thing.)
You are correct, cost is not a big factor. The capacitors ranging from 47uF to 120uF are not significantly different in price ('significantly' being the key word).
..Todd
Johan-Points well taken!
When modeling caps as low as 1u for the last cap (at the 288V B+), I did not consider the effect of the music signal acting "upstream" (still learning!), but I can see it on my scope on another AB1 amp for sure (and it has a 47u cap in that position).
I think there is also confusion about the callouts; the C1-C4 I was referring to are on the PSUD model, not the schematic. The schematic lists C1-C6.
As far as eliminating C4/R1, would that eliminate the (small amount of) decoupling from the output tube B+? I thought it was not good practice to take the small signal stage B+ from the same tap as the outputs....
Last edited:
I think it would be benefiticial to post the latest version of the schematic. I think we are on TAJ version 7, but not sure anymore.
I too thought it was important to decouple each stage of the amplification process. The most important being decoupling the input/driver section from the output. In our case here, the last section of the power supply for the input tube was mostly a voltage reduction step and the cap only really served a purpose for shunting HF noise to ground. I am interested in learning if this was a wrong assumption.
I too thought it was important to decouple each stage of the amplification process. The most important being decoupling the input/driver section from the output. In our case here, the last section of the power supply for the input tube was mostly a voltage reduction step and the cap only really served a purpose for shunting HF noise to ground. I am interested in learning if this was a wrong assumption.