Help with HT power supply

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

earthsled

Well-known member
Joined
Dec 1, 2009
Messages
405
Location
Asheville, NC USA
Hello all!

I have a power transformer with a secondary rated for 240V AC at 125mA.

My mic pre requires 290V DC at about 25mA per channel.

Admittedly, 240V AC is not an ideal voltage for deriving 290V DC. I'm not sure it's even possible to make a power supply for my needs using this transformer.

I found this online, but I haven't tested it yet...
the-variable-high-voltage-power-supply-0-300V.jpg


Would the circuit (above) be suitable for supplying HT voltage to preamp tubes?

Thanks for the help!
 
I haven't got enough knowledge for judging that circuit, but what I would do if I were you, just get the appropriate power transformer. 

i understand that size can be an issue though it shouldn't change much at those values, but wait for someone else advice first
 
earthsled said:
I have a power transformer with a secondary rated for 240V AC at 125mA.
My mic pre requires 290V DC at about 25mA per channel.

Admittedly, 240V AC is not an ideal voltage for deriving 290V DC. I'm not sure it's even possible to make a power supply for my needs using this transformer.
Why not? It looks just fine.
I found this online, but I haven't tested it yet...
the-variable-high-voltage-power-supply-0-300V.jpg

Would the circuit (above) be suitable for supplying HT voltage to preamp tubes?

Yes, as is, but to get optimal results it should be tweaked a bit.
 
The transformer seems fine. I'd like to notice that circuit doesn't regulate the voltage, just divide it. So with different loads or with different main voltage you'll have different voltage. Probably not a big deal, the circuit will probably work fine with the expected ±~10% in some bad conditions.

I'd put a cap in the gate as well to reduce the noise, taking advantage you have the MOS already there.
 
Thanks for the comments, everyone!

I'm happy that using a 240V AC transformer for 290V DC seems feasible. If the MOSFET circuit is the best option, I'm happy to give it a try. It would be ideal to have some regulation, but I don't think its absolutely necessary (as joaquins suggests). The MOSFET circuit is adjustable, but I don't really need this feature either.

The transformer in question is actually from a Fender Blues Jr. guitar amp. I like the PT because it also has a 20V secondary that can be used to power op-amps, LEDs, phantom, etc. (Not bad for off-the-shelf). For this discussion, I have attached the Blues Jr. HT supply schematic. I'm looking at it because TP27 voltage "Z" is pretty close to the DC voltage I need.

Could a simpler RC-type linear power supply be designed to satisfy the given voltages, or is the MOSFET supply the best option in my particular case?
 

Attachments

  • blues-jr-ht-psu.jpg
    blues-jr-ht-psu.jpg
    111.9 KB · Views: 27
You could use the RC filter alone, after the rectifier and a cap, but you have to see what is the resistor you need for it.

240VAC rectified is about 334V after the diodes. At 50mA (for 2 channels) you need to use a resistor to drop those 44V, 880Ω is the number, good luck to get one of those, probably a 1K resistor is your choice. About 2W wasted in it, so use at least a 3W resistor. Even better and cheaper, two 2W resistors, 470Ω for each and you are closer to the value, easier and cheaper. At least here, pretty easy to find 2W MF resistors of any value, not so much for 3W, they are usually WW so not so cheap and much bigger.

JS
 
> Could a simpler RC-type linear power supply be designed

Worked for literally millions of tube systems.

I mean, where would be be if tubes had to wait for the MOSFET to be developed?

And you hardly ever see an R-C filter blown-up by static charge.
 
This is great help!  :)

240VAC rectified is about 334V after the diodes. At 50mA (for 2 channels) you need to use a resistor to drop those 44V, 880Ω is the number, good luck to get one of those, probably a 1K resistor is your choice.
Thanks for your help with the resistor value! May I ask how you are calculating 880Ω?  Four 220Ω resistors in series might also work well here. How critical is the 880Ω value?

Also, how do I calculate the capacitor value?

Thanks!
 
First, you may need some experimentation, I just took your 25mA for channel and assumed 2 channels. If the preamps are not using 25mA the voltage will change, so... Once you get your 50mA for the supply it's just ohms law, 44V at 50mA it's 880Ω. (44/0.05=880)

Here some answers and notes about the details, feel free to stop reading when you please.

About how critical this value is, I'd assume here that the current is constant for this circuit, if I have 300V or 280V feeding it it shouldn't change much at all I guess, that would be the case for most tube circuits. So the error introduced by the resistor is directly proportional to with those 44V. 10% out of the mark for the resistor would mean 4V out for the PS HV. Less than many other sources of error here, that's why I said 1k will be fine. The option with 2x470 is more to use 2W resistors rather than getting a closer value (could be 2x 1k8 in parallel, closer value)

Another note here about component selection and tolerances. It doesn't make much of a difference trying to get closer to the number, the math will throw an arbitrary number, could been 892.9756Ω. So you pick between the ones in the shelf, depending on the situation you may prefer to go to a higher one or a lower one, or the closer one. If they make the jump from 8.2 to 10 in the resistors values is for a reason, E12 in this case, for 10% parts. Let's go and buy 2 resistors, 1k and 820Ω for example. You know it can't be perfect but you'd expect the 1k to be bigger than the 820Ω. So, 820*1.1=909 and 1k/1.1=909.09. As you see here even if you pick the highest 820 resistor and the lowest 1k resistor in a 10% span the 1k will still be higher. This is simplifying quite a bit the situation, You'll never see one that it's just on the border, because it has to meet the specs over a range of conditions. ±5% is fairly typical for a ±10% part at room temp, normal conditions. If I'd choose 1k and 1k2 and do the same there would be a small overlap, but it's fine, and doesn't make any sense to have 3 digit to express something that may be off by a 10%. You don't say this should be 10.2 but could be between 9 and 11.5, it doesn't make any sense. With the time you make those steps automatically, when you are calculating a resistor you'll need to buy and the calculator says 41.41245k you just read 39k. Of course there are cases that need 0.01% resistors and those came in many more different values and flavours but you'll know when you are getting there, let's hope sooner than you imagine!

Good luck and ask as much as you need, that's what this is for.

JS
 
> How critical is the 880 value?

How critical is your "290V"? How sure is your "25mA"?

IMHO, a "290V" tube preamp should work very well from 240V to 350V. Tubes are not fussy.

The "25mA" will vary with B+ voltage. Also with different tubes. And with all other resistor tolerances.

A straight R-C filter will want 3 or 4 stages to clean-up to preamp level.

Best filter economics is all resistors the same, but a little off isn't much different, and DIY is not about best-profit economics.

I'll accept JS' "880" value, because it sounds reasonable, and I'm too lazy to re-do your math for you.

I'd suggest a handful of 500 Ohm and a handful of 1K resistors, perhaps 5W each. Start with a string of 1K with caps. That will probably land below 290V. The preamp may work fine. If you are measuring THD, higher B+ means slightly lower numbers, so you replace 1 or 2 of the 1K with 500 Ohms to bring B+ up. Takes very few iterations to get within 20%, and in tube-work +/-20% is spot-on for all practical purpose.
 
PRR is right. Tubes are incredibly tolerant of HT voltage. There is generally little if any need for stabilised supplies or exact voltages in most preamplifiers. A series of RC filters will provide a means of dropping the raw HT to close to the required voltage and  at the same time reduce r ipple to a suitably low level. The raw HT will be close to the peak value of the ac voltage and a simple application of ohms law will provide the values of the series resistance required. Three equal RC stages will will typically reduce ripple and noise by 90dB - better than any regulator will do.

Cheers

Ian
 
post the mic pre schematic,

copy the V72 supply,

humm will be your main issue, so layout and tube heater circuit is just as important as pwr sup circuit,
 
This information is greatly helpful! Thanks everyone!
I've attached a first draft.

I chose a 3-stage filter simply because I had two 470Ω 5W resistors in the shed. What are the advantages/disadvantages of a 4-stage filter? Are inductors/chokes worthwhile in this case?

Looking closer at the paperwork for this preamp (REDD.47), the manual shows a current rating of 30mA per channel. I was estimating 25mA by adding the currents shown on the schematic. The resistance needed for two channels drawing 30mA calculates to approximately 750Ω (rather than 1k). I'm not sure if this is significant.

Approximating the capacitance is still a bit of a mystery to me too, so I chose 10uF as a starting point.

 

Attachments

  • 290V-HT-PSU-v1.png
    290V-HT-PSU-v1.png
    26.3 KB · Views: 22
I would increase the values of capacitance significantly - 220uF would be about right and possibly add one more RC stage - just change the resistors to 330 ohms.

Edit: As an aside it is interesting to compare the attenuation of 3 RC stages versus a single one. The attenuation of a stage is approximately 2*pi*f*R*C. A single stage of 990 ohms and 660uF has an attenuation of just under 54dB at 120Hz. If instead we use three RC stages each of 330 ohms and 220uF then each stage attenuates by just under 34dB. Three of these in series produces a total attenuation of over 100dB. Same series resistance, same voltage drop, same total capacitance but 70dB more ripple reduction.

Cheers

Ian
 
Ditto Ian's suggestion.

Look at the RCA BA-2 schematic for comparison.  Also note that the 40uf caps they show would have probably been much higher due to typical tolerance of the caps for that era (call them 60-80uf ea).  The BA-2 draws just about 4 ma total, which is four times less per channel than the version of the REDD you're building 
 
Excellent!
I've attached a second draft.

Looks like I'll need to allow some space for these big caps!
Should I be concerned with inrush current or anything like that?
Do I need an additional reservoir cap after the bridge?

What about a shunt resistor(s) to drain off the voltage when the unit is powered off?
The REDD.47 PSU had one or two 470kΩ which where used for this, as I recall.

Also, I'm curious about using small caps either on or before the bridge.
(The Fender Blues Jr.  schematic has a 0.1uF on the transformer secondary).
I've seen this done before, just not sure if it's necessary.

Thanks again!
 

Attachments

  • 290V-HT-PSU-v2.png
    290V-HT-PSU-v2.png
    27.4 KB · Views: 22
You NEED a Cap right off the rectifier to make raw DC.

While a resistor in front of the first cap will "work", output voltage will fall way off (your "334V" note ass-umes a cap there!) and the heat in that resistor is hard to compute. Dozens of Ohms might be used here to limit rectifier peak current, but with today's rects and preamp loads, that's not needed.

> What are the advantages/disadvantages of a 4-stage filter?

My well-water is brown and gritty. One filter takes the rocks and grit out, but runs brown in the bathtub (uck). Another filter leaves it nearly-clear in the tub, but if you fill a bottle and let it set there's some brown settles out. A third filter reduces the visible brown to nil, but is it safe enough for the infant to drink?

A single filter to take-down deep-brown to near-clear may be possible, but likely much more expensive.

Also you typically have several needs with different cleanliness requirements. I take my garden hose after the first filter. The grit is bad for the spigot, but a little brown won't hurt the grass, and the second filter does not have to process this water, works better.

> Approximating the capacitance is still a bit of a mystery to me

Can you figure a High-Pass filter?

What would you use for coupling a 500 Ohm load? Take 20Hz. I would pencil 20uFd. Check: 20uFd and 500r is 15.9Hz.

Here we have a low-pass at 16hz, fed 120Hz crap. Crap reduction is roughly 16/120 or nearly 8:1. A good reduction but not fabulous. Certainly not enough to turn raw DC into smooth-enough for preamp DC.

> compare the attenuation of 3 RC stages versus a single one.

Just looking at 120Hz, it is about (1/8)*(1/8)*(1/8) or 1/512. That's a lot better, at 120Hz.

But the reduction for higher overtones of 120Hz (the "buzz") is much better. At 240Hz one stage is 1/16, three stages is 1/4,000!

Economics also comes into play. One stage requires one BIG cap. 4 stages means 4 smaller caps, and at some point you get a price-break for ordering 10 parts.

And one BIG cap is NOT a good plan, because the absurdly low calculated impedance is foiled by wiring resistance.

My rule of thumb is to take 20dB (10:1) to 30dB (31:1) in each stage. This means picking low-pass corner as 12Hz to 4Hz.

Before that you MUST get a rough estimate of raw-DC crap (2%-5%) and how clean the end-filter output must be. If this is a PA power-amp you may tolerate 10mV of crap fed to the first stage plate. If it is a mike-amp, you want to be well below 10 MICRO-volts at the first grid, which for typical first-stage gain is 300uV (0.3mV) at the feed to the first plate. Transistors have to put DC to the first Base so may need a much cleaner node (albeit at extremely low current). Many-transistor amps may have so many current-sources that a part-Volt of crap doesn't corrupt signal. Understand what you are building.

> Are inductors/chokes worthwhile

A resistor is the same impedance for 120Hz or DC. A choke has much higher impedance for 120hz than DC. So you get more crap-filtering with less DC drop.

This can be valuable when the DC load is large, or varying. You often find a screen choke in a guitar amp, because they like their screens at high voltage, but screen needs to be 5X cleaner than Plate, and in hard work the screen current varies a lot.

Preamps are nearly constant-current. And in today's market, an over-Volts PT is cheaper than a right-Volts PT plus a second core for a choke. Look around, you find nearly no commercial or professional preamp power supplies with chokes. (Pre-1940 may be the exception: large caps were more expensive than dumb iron in those days, and what was spent on the choke might be saved on the caps. Today e-caps are dirt-cheap and chokes are all specialty items.)
 
As PRR says, you need the (reservoir) capacitor right after the bridge to create the raw dc. There is a series resistance due to the rectifiers and the secondary resistance of the transformer. The size of the ripple on the raw dc is determined by the size of this capacitor and your dc current draw. They are related by the classic formula:

i = C*dV/dt

To  first approximation, i is your dc current, dV is the peak to peak ripple and dt is the period of the waveform (8.3mS for 120Hz). The subsequent RC low pass filters reduce this ripple to an acceptable figure. Rearrange this equation to be more useful as:

dV -=i*dt/C

which allows us to calculate the approximate ripple. With C = 220uF, dt = 8.3mS and i = 25 mA we get a ripple of just under 1V peak to peak. To get this below one microvolt you need 60dB of smoothing. If C was 22uF you would have nearly 10V ripple and need an extra 20dB of ripple reduction. It is tempting to make the reservoir capacitor as large as possible but the larger it is the greater the diode current and the lower the transformer efficiency but this is a whole topic on its own.


Lastly, you do need a snubber capacitor across the bridge to kill and spikes cause when the bridge diodes turn off. These have a nasty habit of finding there way into the signal chain. I use a 100nF 275VAC film type across the ac input to the bridge. Others prefer a 100nF acros each rectifier but I have not found this to be necessary. You can reduce turn off spikes by using fast recovery diodes e.g UF4007

Cheers

Ian
 
I'm so grateful for this discussion and your assistance!
Thanks everyone!

> You NEED a Cap right off the rectifier to make raw DC.
My apologies for not realizing the necessity of the first "reservoir" cap.  :-[
This is my first attempt with creating a linear power supply from scratch.
I obviously have a lot to learn!

> With C = 220uF, dt = 8.3mS and i = 25 mA we get a ripple of just under 1V peak to peak.
This makes sense to me, but shouldn't I be calculating for 50-60mA current?

> I use a 100nF 275VAC film type across the ac input to the bridge.
Fender used a 630V cap here, so I assume that's an appropriate voltage rating for this transformer.

Third draft attached.
Please let me know if I'm headed in the right direction!

 

Attachments

  • 290V-HT-PSU-v3.png
    290V-HT-PSU-v3.png
    26.1 KB · Views: 14
Assuming 60 ma is the correct current draw, as you have it drawn your resistors should be about 250 ohm (instead of 330) for the given voltage drop.  One thing to remember is that you're chasing a moving target - as you raise or lower supply via dropping resistors, the current draw changes.  Overall, yes you're on the right track and probably close enough.  With 330 0hm X 3 @ 60 ma you will end up with final B + of 274-275V

As many have already said, tubes aren't fussy and the circuit will work fine, but if final B+ voltage needs to be within a certain range for the proper bias of the output driver, sometimes the best way is to put it on the breadboard.  It's probably hairsplitting for some small percent increase or decrease in total THD.  But just so you have some techniques at your disposal, putting the PSU on a breadboard and substituting the resistors until you get the final voltage you aimed for is very good for understanding how all the paper theory works. 
 

Latest posts

Back
Top