High Voltage Toggle Switch Suggestions

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

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

millzners

Well-known member
Joined
May 7, 2010
Messages
219
I'm adding a switch or switches to my tube pre to change the HT voltage in order to introduce some distortion via starved plate.  I found out accidentally that I liked the way the tubes sound under this condition, but I've never been able to find a suitable switch in my inventory of stuff.

Kingston does the same thing I want to do with his Drive 1, and recommends a reed switch relay from digikey:  306-1195-ND http://www.digikey.com/product-detail/en/9104-12-10/306-1195-ND/1154390  which is great, but I don't have 12VDC I can use for the control voltage and I'd rather not add an additional supply rail just for that one thing.

I found these:  Carling DK284-73 http://www.mouser.com/ProductDetail/Carling-Technologies/DK284-73/?qs=%2fha2pyFaduhLFihSPIXLvK6XeHtcP7imEE1LdKwPvBYzwukJY%252bgnMg%3d%3d which says 250VDC but I don't think that's nearly enough... 

So I thought I'd ask to see if anyone has any suggestions.  Obvious a schematic would be helpful, but until I find the schematic for my build but here is Kingston's where SW6 in the power supply is pretty much what I had in mind:  http://www.michaelkingston.fi/files/Drive-1_schematic_rev1.9.pdf

Bonus points if you know of a rotary switch that can handle this task.
 
while that grayhill will work, it is not rated to switch 600vac from what I can gather in the data sheet. their test data is 1k Hz 115vac and 30vdc.
these are somewhat standard voltage numbers for testing purposes/certifications.
The options are limited for something officially rated for your application. But many parts will work.
I would look for a solidly built non-shorting ceramic switch on the bay/surplus, physically bigger than that grayhill.
 
250V rating is fine for 320V switching in a mike preamp. Current is very low, and you own a soldering iron. If rated life is 5,000 operations, and 320V kills it in 2,000 operations.... how many sessions is that, and how long will it take you to replace it?

Don't hard-switch large (47uFd) caps. Have some resistance in the loop.
 
build the Fairchild 670 regulated pwr supply,

no switches, just a knob on a pot that controls voltage,

that way you can exactly dial in the sound you want,

goes from 285 down to about 100,

even lower if you change a resistor value,
 
i use an adjustable hv psu for all my tube circuit testing. it uses a pot to dail  in volts from 30v to 400v dc. it was a kit i ordered from hong kong, and it not bad quality.
ill have alook if i can find a link to the site or a schematic.


 
Thank you for the replies.  I love the idea of a knob. 

How about this in series with the HT coming off of my 5Y3G rectifier tube before the choke (wired as a rheostat)

http://www.digikey.com/product-detail/en/026TB32R103B1A1/CT2159-ND/203780

5W
500VDC
10kohm
$4.63 (DIRT cheap)

I measured roughly 6mA per channel a while back (without sending it a signal), so ~20mA for 3 channels across 10kohm would give 200V drop.  If I remember correctly I have somewhere in the neighborhood of 400VDC coming off my rectifier tube so after the other subsequent resistors that should get me in the ballpark of 80VDC supplying the NYD 6SN7 circuit... 

I had a senior engineer tell me once that using a pot in this way was a bad idea and I should use a rotary switch instead which was too much hassle at the time. 
 
leadbreath said:
i use an adjustable hv psu for all my tube circuit testing. it uses a pot to dail  in volts from 30v to 400v dc. it was a kit i ordered from hong kong, and it not bad quality.
ill have alook if i can find a link to the site or a schematic.

I wonder if the knob in your device is a Variac before the step up xfo and rectification, which was another solution I considered since I have a couple Variacs laying around until I read that this is very bad for tube rectifiers to get much less than 110VAC.  I guess that would be an expensive way to do it on second thought...
 
> I read that this is very bad for tube rectifiers to get much less than 110VAC

??

For a 400V supply we typically feed 300V AC. Not 110.

OTOH we have rectifiers fed 3V or less for signal detection or control.

You *must* keep the *heater* voltage on-target +/-10%. So to get any real change of B+ you must run a separate heater transformer not on the Variac.

> ~20mA for 3 channels across 10kohm would give 200V drop.

Except it won't still suck 20mA at the lower voltage.

Figure the preamp as a large resistor. At 400V it draws 20mA, so it is "like" a 20K resistor. Put 10K in series, it drops to 2/3rd, or 267V. Not a lot of drop.

In many cases it would be simpler to leave the B+ alone, crank the gain going in, then pad-down the audio output 2/3, 1/2, even 1/10th, to get your dirt without blowing-up the next box in the chain. Padding fractional-Watt audio is simpler (and cooler) than dumping whole Watts in the power supply.

> a senior engineer tell me once that using a pot in this way was a bad idea

Yeah, well, senior engineers get these ideas. He's not 100% wrong. In extreme cases using part of a pot will burn-out that part because it can't shed the heat that the full pot would stand. That's not so true when the load is a fairly large resistance: you won't get a huge current in a small part of the pot. Wiper resistance is an issue but a true 5W pot (not a re-rated 2W) is made for large wiper current. Wiper reliability is always an issue.... if the wiper hits dust and opens the circuit, does anybody die? Not in this case, though loss of B+ even momentary  could ruin the best take of the night.

The real problem is that it takes a HUGE range of series resistance to get a really significant change of voltage. 2K will put you near 360V, 20K for 200V, and 80K for 100V. So if you wish to get down into the truly low voltages which may be musically different, then up at the higher voltages the pot will be very twitchy.

Voltage divider is smoother but must dissipate much more than the load.

Hi-R divider with cathode follower (these days, a MOSFET) is more efficient. Such things are sold to guitar-amp hackers under names like VVR and Power Scaling(R). Your 20mA load is a loaf for a Champ-size VVR.
 
PRR said:
> I read that this is very bad for tube rectifiers to get much less than 110VAC

??

For a 400V supply we typically feed 300V AC. Not 110.

OTOH we have rectifiers fed 3V or less for signal detection or control.

You *must* keep the *heater* voltage on-target +/-10%. So to get any real change of B+ you must run a separate heater transformer not on the Variac.

I'm embarrassed I didn't think that through before I typed it, yes that's what I meant.  If I had the room I might consider a variac option by putting my heaters on another xfo I have laying around.  I'm tempted to do this because I have all the parts I need to do it, but the variacs I have are probably too big for my enclosure.

> ~20mA for 3 channels across 10kohm would give 200V drop.

Except it won't still suck 20mA at the lower voltage.

Figure the preamp as a large resistor. At 400V it draws 20mA, so it is "like" a 20K resistor. Put 10K in series, it drops to 2/3rd, or 267V. Not a lot of drop.

In many cases it would be simpler to leave the B+ alone, crank the gain going in, then pad-down the audio output 2/3, 1/2, even 1/10th, to get your dirt without blowing-up the next box in the chain. Padding fractional-Watt audio is simpler (and cooler) than dumping whole Watts in the power supply.

I wish I could do this but the combination of rat's nest PtP and limited space means I'd rather leave that part of the circuit alone.

> a senior engineer tell me once that using a pot in this way was a bad idea

Yeah, well, senior engineers get these ideas. He's not 100% wrong. In extreme cases using part of a pot will burn-out that part because it can't shed the heat that the full pot would stand. That's not so true when the load is a fairly large resistance: you won't get a huge current in a small part of the pot. Wiper resistance is an issue but a true 5W pot (not a re-rated 2W) is made for large wiper current. Wiper reliability is always an issue.... if the wiper hits dust and opens the circuit, does anybody die? Not in this case, though loss of B+ even momentary  could ruin the best take of the night.

He thought intermittent wiper contact would hurt the tubes.  I believe you if you tell me it won't, he hadn't seen a tube in years.

The real problem is that it takes a HUGE range of series resistance to get a really significant change of voltage. 2K will put you near 360V, 20K for 200V, and 80K for 100V. So if you wish to get down into the truly low voltages which may be musically different, then up at the higher voltages the pot will be very twitchy.

Voltage divider is smoother but must dissipate much more than the load.

Hi-R divider with cathode follower (these days, a MOSFET) is more efficient. Such things are sold to guitar-amp hackers under names like VVR and Power Scaling(R). Your 20mA load is a loaf for a Champ-size VVR.

Seems like the VVR concept is exactly what I should be after, in that it is placed where I am intending to place my switch/knob and it does exactly what I'm intending my switch/knob to do.  I think I'll go looking for a schematic.  Thank you so much for your advice.
 
> He thought intermittent wiper contact would hurt the tubes.

Then your "senior" engineer isn't senior enough; or as you say, is out of practice.

We routinely put a tap-switch in series with a tube to tap out Morse Code. di-dah-di-dah-dah-dah all night long. The problem with this is that hard-keying a radio oscillator makes an ugly signal. Do this with a KiloWatt rig and you get complaints from all up and down the band. But widely used for code practice oscillators and flea-power transmitters.

All AM/FM radios broke the B+ to the tubes not used for the mode selected, AM or FM.

Breaking the -heater- frequently is an issue, though minor. Like incandescent lamps usually blow at turn-on. Tube heaters are good for a few thousand turn-ons. In most uses you run through your 5,000 hours of steady-on life before you accumulate too many heater turn-ons.

But you are not proposing (shouldn't) switching heaters.

If you needed to, there are several fast-heat tubes aimed at mobile taxi/police/etc, which could be lit-up when the used pressed the mike-switch and left cold between, many times a shift.
 
I've cobbled together the attached schematic (http://yeomansinstruments.blogspot.com/2008/07/vvr-greatest-thing-ever.html)

This is my first time using a MOSFET to regulate a power supply, but going back to my days in school I can see it's a voltage follower circuit where the pot adjusts the reference voltage on the gate which changes the output voltage.  But I'm unclear on what the zener does, other than pass current if the gate to output voltage wants to be greater than 6V.  I'm curious because I only have a 15V zener on hand and I don't know what effect that will have on the circuit.  I'm just digging around my stash for parts here (the MOSFET was taken out of a PC PSU) and so I was hoping to use what I have. 

I've also scaled back the power rating of the 10R.  I'm at about 20mA and, for example, the 5W 10R can be reduced to a 1/2W 10R which will still handle 220mA.
 

Attachments

  • VVR in 18watt schem.png
    VVR in 18watt schem.png
    30.7 KB · Views: 13
I found a zener that measures 9.5V so I may just give it the 'ol college try.

Edit:

Finally found some info on the reason "why" 6.2V, and I read that 6.2V and below rated zeners are quieter than above 6.2V while zeners less than 5V don't regulate very well.  And while I tend to be a little skeptical of things I read on the internet it does seem to be a very common value when looking at these regulator circuits.

I wish I could stick my spectrum analyzer on my B+ to see a before and after.

Edit 2:

Found this:

"The addition of a current limiting resistor will save the FET. R3, 10 ohm in the next diagram, is the current sensing resistor. With no current, there's no drop across R3. Thus, the voltage across the 6.2V zener diode, D1, will only be 3V (the FET threshold voltage). However, when the load current reaches 300mA, R3 is dropping 3V. Add that to the FET threshold of 3V gives 6V across the zener, which is enough to start it conducting. If the load increases beyond 300mA, the zener will pull down on the gate and thus reduce the output voltage. The R3/D1 circuit essentially will not allow the FET to conduct more than 300mA."

from this site http://paulrubyamps.com/PSMDesign.html, but I changed the reference designators and values to match the design I'm working with.

Ok so now if I apply that logic, and remember that at no time would I expect this circuit to pull more than 50mA, and I'm stuck with a 9.5V zener, then the values would work out to about R3= 130R.  That would give a 6.5V drop with 50mA and the 6.5V + the Vgs of 3V would give me some over current protection.
 
Here's a recent thing for generic high voltage duties and this exact starved plate testing in mind: http://www.groupdiy.com/index.php?topic=51232.msg651854#msg651854

Use the zener to set your highest voltage and the 1meg pot (or a rotary switch) sets the drop. Let the MOSFET take all the heat. Caps can be electrolytic even if I've marked then film.

I wouldn't use the (reed) switch and a big-watt dumb resistor thing anymore. It was a hack for a time I had already etched a PCB and had very little space for mods.
 
Kingston said:
Here's a recent thing for generic high voltage duties and this exact starved plate testing in mind: http://www.groupdiy.com/index.php?topic=51232.msg651854#msg651854

Use the zener to set your highest voltage and the 1meg pot (or a rotary switch) sets the drop. Let the MOSFET take all the heat. Caps can be electrolytic even if I've marked then film.

I wouldn't use the (reed) switch and a big-watt dumb resistor thing anymore. It was a hack for a time I had already etched a PCB and had very little space for mods.

Thanks for the reply and the link, I've also been studying your schematic for the M670 especially the power supply as it relates to this situation.  That's another awesome project that I'm looking forward to building.  Looks like based on your link and what I've got the only thing missing is the first zener to limit the max voltage which is fine since I'm cool with the ~300V I'm getting now.  Since I ripped the mosfet out of a PC power supply it came with a good heat sink ready for mounting to my enclosure.

CJ said:
try grid leak bias, nice compression, a little noisy but parts count is low, (1 cap, 1 resistor)

Would you recommend the series or shunt configuration?  This may be out of reach for this project as I'm using 6SN7GT and I'm worried it doesn't have high enough mu (spec says 20).  It also specifically says on the datasheet to use a cathode bias resistor or other "adequate means" to protect the tube when grid-leak biasing.  Seems like something to use in my next project which will have 12AX7 on the input, so thanks for the suggestion.
 
I wired up the circuit and tested the output voltage.  Typical voltage after the choke (!) was 300V, now voltage ranges from 300V to 40V on the output of the VVR.  So I'm almost there.  There is some wild noises that it makes now; on power on it sounds great for a little while then it makes clicks, then if you play around with the new knob a little bit it does the staved plate but after a minute or two it goes into this distortion that sounds like an octavia (untuned harmonics, aka oscillations) that affects all 3 channels and occurs even when the plate voltage is set back to 300V and won't go away until the thing is powered down and back on. 

So in the RF world we call that "spurious" which is a condition of instability usually caused by oscillating power supplies, opamps, or chip amps.  Unfortunately in this case I can't apply the 'ol magic finger (literally my finger) and then check with my scope and analyzer to see where it's coming from.  Anyway I'm dubious of the choke, and of the lack of any cap on the input to the MOSFET... 
 
Did you forget the resistor to gate? "zero" impedance at the gate causes exactly the type of thing you describe. It was in the thread linked earlier. Anything around 220-1k ohm works.
 
Kingston said:
Did you forget the resistor to gate? "zero" impedance at the gate causes exactly the type of thing you describe. It was in the thread linked earlier. Anything around 220-1k ohm works.

You're awesome.  Yes, I also excluded the 10uF cap on the gate.  Once I put a 221R and a 10uF cap on the gate it purrs like a kitten.  Thank you!
 
Back
Top