Low-noise power supply without a choke

[thorpej]

Diversions from day-job work desperately needed, so... been playing with the power supply for the 6550 amp I'm contemplating.

I'm concerned with using a choke in the power supply because of the inductance drop at full power. Even the 3H Hammond 193N is only rated at 500mA. I used this choke with much success in my 30W/ch ultra-linear 6L6GC amp, but my 6550 design has the output stage (total for both channels) idling at 300mA. So I decided to toy with the idea of a no-choke filter.

Nichicon makes some high-voltage, high-capacity electrolytics with reasonable ripple current rating that seem like a good candidate for this application. My favorite bridge rectifier has a ridiculous-A surge and 35A steady-state current rating, so I'm not particularly concerned about in-rush current.

Power transformer is a 400VA AnTek toroidal with a 360V secondary.

I decided to use a 1000uF @ 525VDC reservoir cap, followed by a 51R / 1000uF @ 525VDC filter. The filter is a low-pass filter at ~3Hz. Simulation shows in-rush is about 27A, which is well within the ratings of my chosen rectifier. I'm aiming t have 450V available at the plates of the 6550s, and expect a 3V drop on each half of the output transformer. The circuit as-drawn simulates at 455VDC at the OT center tap with a 42mV ripple.

I know @BinaryMike is an advocate of choke-less supplies (they don't ring when you hit demanding transients). Would love to hear others' feedback on this.

 

[whoaru99]

Last amp I built doesn't have a choke for the plate supply for similar reason. In fact, I came right off the first capacitor. I used a small choke for LTP / front end though.

 

[thorpej]

Last amp I built doesn't have a choke for the plate supply for similar reason. In fact, I came right off the first capacitor. I used a small choke for LTP / front end though.

Simulation shows pancake-flat supplies for LTP and front-end as-drawn; I'm not really concerned with those right now, though I could consider it.

I like to have a bit of extra filtering for U-L topologies, and not rely on common-mode noise rejection inside the OT, hence not coming off the first cap.

 

[whoaru99]

I used 500uf capacitance at the C1 position so a small (but overkill for just the front end) 120mA choke already on hand worked well downstream to get the smoothing.

 

[whoaru99]

I like to have a bit of extra filtering for U-L topologies, and not rely on common-mode noise rejection inside the OT, hence not coming off the first cap.

I understand. I did it in part for experimentation just to see, with some expectation I might've had to revamp it but I didn't (need to).

 

[kward]

Even a much smaller inductance choke with higher current capability I think would ripple reject better than a 51 ohm resistor. Of course you've got a whopping 1000 uF behind it so that may be just as good.

But just for kicks, you might want to try a simulation run with a CLC of 100 uf, 0.6H, and 500 uF, and use the Hammond 159Y or something similar to it (750 mA current capacity). If it rings, you could make it a RCLC, with the R being a wirewound 15 watt type of maybe 10 to 30 ohms.

Just an idea anyway. The 159Y is small enough where it could be mounted under the hood, if you're using a 3" high chassis.

 

[kward]

Another approach is ripple rejection with a pseudo HV regulator consisting of a mosfet with a zener regulated gate. Pete Millett did something like that on one of his amps (the DCPP "Engineer's" amp). It would require heat sinking the mosfet. Not advocating this approach necessarily, just noting that someone tried it.

 

[Bob]

you can try the lr8n3 - high voltage regulator up to 480v - good regulation (by my standards)
but only passes 20ma so you'll need a booster transistor to get past 20ma.

if you want one - PM me I'll send it for free.

another method I like is a series pass regulator using tubes like the 6080 (from memory)
and was going to do an EL34 version based on the theory that worn-out EL34s came be moved
from the power side to the PS side, with multiple EL34s in parallel to provide more current.

http://www.mouser.com/ProductDetail...=sGAEpiMZZMtUqDgmOWBjgKyTpCrAaldB3QI3ZKIcezc=

 

[gadget73]

Would splitting the supply per channel to allow the use of commonly available chokes be an option? That would halve your current demand, though of course it basically doubles the component count in the B+ supply so space may become a factor.

Honestly I get the appeal of mono amps once you get to a particular power level. Makes designing it a bit easier.

 

[vinylkid58]

Another approach is ripple rejection with a pseudo HV regulator consisting of a mosfet with a zener regulated gate. Pete Millett did something like that on one of his amps (the DCPP "Engineer's" amp). It would require heat sinking the mosfet. Not advocating this approach necessarily, just noting that someone tried it.

It works quite well. I've talked to designers that use this approach with SS amps also.

jeff

 

[Bob]

there are cheap boards on eBay that do this HV "regulator" thing. it would be a cheap and
easy way to decide whether to use or not. some are tube based rectifier PS, some are high voltage
(250V!!) regulators but can be modified for higher voltages. they are usually between 5-10 USD.

 

[thorpej]

Even a much smaller inductance choke with higher current capability I think would ripple reject better than a 51 ohm resistor. Of course you've got a whopping 1000 uF behind it so that may be just as good.

But just for kicks, you might want to try a simulation run with a CLC of 100 uf, 0.6H, and 500 uF, and use the Hammond 159Y or something similar to it (750 mA current capacity). If it rings, you could make it a RCLC, with the R being a wirewound 15 watt type of maybe 10 to 30 ohms.

Just an idea anyway. The 159Y is small enough where it could be mounted under the hood, if you're using a 3" high chassis.

Well, indeed, simulation shows even a 0.6H choke is way better at this. I put 110uF in front of it and 680uF after, and it's pretty smooth, with a higher voltage output - 468V - (which I guess I'll have to tamp down a bit given the 6550A is only rated at 450V max for U-L operation, according to the GE data sheet). And even with the choke, it appears to cost less (those 1000uF caps were ~$42 a piece).

 

[thorpej]

Would splitting the supply per channel to allow the use of commonly available chokes be an option? That would halve your current demand, though of course it basically doubles the component count in the B+ supply so space may become a factor.

Honestly I get the appeal of mono amps once you get to a particular power level. Makes designing it a bit easier.

Yah, doubling the component count makes it pretty tight. I'm already bad at keeping things neat in tight spaces, so I don't want to go that route I considered a split plate supply with 2 chokes on the 6L6 version of this amp that I built, but ended up using the Hammond 193N -- physically large, but I got it all to fit top-side, even with that toroid-in-a-can PT.

 

[knockbill]

I like those big Nichicon reservoir caps you used,,, got a part number? Don't think I'd need 1000uF,,, but if you have the series or model number, I could look em up..
Thanks...

 

[thorpej]

I like those big Nichicon reservoir caps you used,,, got a part number? Don't think I'd need 1000uF,,, but if you have the series or model number, I could look em up..
Thanks...

The part number is shown in the image attached to message #1 -- LNUN7102MSEF. Data sheet here: http://www.mouser.com/ds/2/293/e-lnu-879958.pdf

In the 525VDC rating, they have 680uF, 1000uF, 1500uF, 1800uF, 2200uF, 2700uF, 3300uF, 4700uF, and 5600uF.

 

[knockbill]

Thanks... I must have missed it...

 

[maxhifi]

I like to have a bit of extra filtering for U-L topologies, and not rely on common-mode noise rejection inside the OT, hence not coming off the first cap.

Fair enough, but common mode noise/hum rejection does works great with UL Push-Pull amps. On the UL amplifier I built, I used only a single filter section, and no choke, and there's no issues with hum or noise. On single ended amps though, all bets are off.

 

[primosounds]

I like those big Nichicon reservoir caps you used,,, got a part number? Don't think I'd need 1000uF,,, but if you have the series or model number, I could look em up..
Thanks...

I think with big caps you pay a sonic price. The charge/discharge cycle is slower with big caps. To me this shows up as a lack of fine detail and loss of harmonics in vocal and instrumental presentation. I realize you are not making a SE amp but many of the se diyers don't use caps bigger than 100uf. Even the classic amps seldom had caps bigger than 200uf and those were mostly in voltage doubler power supplies.

 

[PakProtector]

There are quite big charging currents with the filters y'all have put forth. Scope the primary, and see what happens... IME it is useful to keep this stuff out of the heater circuits, and thus the cathode's...which for me are usually one in the same thanks to using DH tubes. A +Critical L-C filter takes much more gentle bites; it is nearly a DC delivery from the power trans. Putting a large cap into one of those is a fine idea, and has worked very well for me.
cheers,
Douglas