T-reg HV regulator
Update October 2019
I have boards and parts available in a Group Buy at diyaudio.com. We will discuss circuitry, construction and what have you there. This place is where I will maintain documentation such as updated schematics, parts lists and board layout.
Update 4 Nov 2019: Some time ago someone wondered what would happen if you connected the output to a fully charged capacitor while the input was off. It's not clear to me how that could happen; where you'd get a fully charged cap unless you have the regulator on. But this is easy to simulate so I did that. What happens is that the fully charged output cap discharges via the pass device parallel diode, which is normally off. I see a current of some 8uA flowing through the opamp inputs into the reference cap. In that scenario the max allowed voltage difference between the opamp inputs is exceeded. Why there is only 8uA is probably because the opamp model doesn't fully model the input stage over-voltage. I made a quick fix to replace D3 with a 2.7V zener diode. That keeps the opamp input stage voltage below some 2.5V, no problem. I just tried it out on my prototype and I see no difference in performance. So, if you are worried about that extremely unlikely scenario, put in a 2.7V zener at D3 and relax. Mouser 625-BZX85C2V7-TAP is a good one.
You have a choice of on-board heatsinks. I used Fisher but Aavid and others have mechanically compatible ones. Depending on the dissipation of your pass device [which of course is (Vout-Vin)*Iload] you should use as shown in the table. For higher dissipation levels, use the Q7 position and mount on large heatsink or on chassis:
Update 28 July 2019:
Just finished the basic tests on the new version. It's become a very nice and compact unit wihich works really well. The new protection system works excellently. You set the maximum current with a two resistor value. When the maximum current is reached, the first stage limits the current to the set value. If this overload continues, the 2nd stage shuts down the regulator after about half a second. You then need to reset the supply by either momentarily removing the input voltage or momentarily remove, then replace, an on-board jumper.
One more change to the PCB and then I'm done!
Update July 2019: It's time for an update to this project, especially now that I have found a way to run it stand-alone, without needing a separate heater winding for the MOSFET version. The older unit needed a floating heater winding which was a nuisance. The new unit will also be much more robust, pretty much undestructable whatever you throw at it. Current limiting of course, and shut down with prolonged overloads and shorts. Specifications adapted to max input voltage of 630V, max output voltage 600V. If necessary it will handle hundreds of mA, even half an amp will be no problem. With the same stellar performance.
I have some PCBs left for this design that are limited to 450V absolute max input, 420V max output, so if you are interested, let me know.
Update 2 Nov 2017: AudioXpress has placed my AX T-Reg article online. Thank you Editor-in-Chief João Martins!
This is a project for a high-voltage regulator for tube amplifiers. The design will provide anything from 50V to 500V DC and more at up to 1A DC. The tube on the assembly in the picture is an EL84 for the preamp 50mA max version. The large 6528 plug-in will provide up to 500mA. The MOSFET regulator plug-in can handle up to 1A output current.
Update November 2016:
I have updated the article but it has not been published. Several people asked me about a PCB for the updated design, but only a few have been produced. To accomodate interest, I have decided to publish the article and the PCB Gerber files here so people can have their own PCB manufactured and/or organise a Group Buy at diyaudio. Here's a color rendition of the layout showing a few SMDcaps at the solder side.
Of interest: I received a message from Jacques Dehaye from France, with a link to "The American Electricians Handbook". This book has a Motorola app note for an MC1466L and its cousin, the MC1566L.This is a clear diagram explaining the operation of the floating regulator. Thank you Jacques!
Also, John Walton alerted me that resistors R3 and R5 should be able to handle high voltage, up to 500V. These are available from the usual sources like Farnell and RS Components, from Multicomp and Vishay. Alternatively, you could make up R3 and R5 from two resistors in series of half the total value.
Stop press! There is an error on the DN2540 plug-in board available through Elektor. One end of R3 is going nowhere - this should be connected to K2. My apologies, it is entirely my fault. Thanks John Lopez!
As noted, the T-reg is based on the MC1466L floating voltage/current reg chip, now obsolete. Interestingly, Gary Lecomte has developed a discrete version (pcb actually) of that MC1466L! (Thanks John O'Neill).
T-reg update - important
One builder of the DN2540 version of T-reg had some problems with it and I discovered that it probably is related to (re)connecting a capacitive load with the regulator live. The pass device DN2540 has a max Id of only 500mA, which is sufficient for normal use but may be exceeded in some cases when (re)connecting a cap load.
Elektor have published the design in the March 2009 issues of the NL and UK magazines. AudioXpress has published an adapted version in their April 2009 issue.
The T-reg Stuffing Guides in the AX-article didn't come out very well so here are alternatives: Main board, 6080 board, EL84 board, DN2450 board.Boards for the original design may be obtained from Elektor through their PCB service - I have no more PCBs for this project available.
Update May 2012: I have redesigned this regulator as a stand-alone board with only the regulator, no rectifier, no time delay. You input the raw rectified DC and a supply voltage for the control circuit, and out comes the regulated voltage. Also redesigned the unit for a 'regular' enhancement mode MOSFET, tested with an IRF740. Works well, almost unlimited current capability, with substantial the same specs as the old unit, with less mains breakthrough. Can be 'hot plugged' into a load with no problems.
Update February 2013: I have submitted a follow-up article to Elektor, describing the new version which uses 'regular' power MOSFETs as pass devices. I expect Elektor to publish it later in 2013. This article also shows how to use a pos reg for a negative voltage and vice versa, which means that the DC supply for the control voltage no longer needs to float on Vout, but can be ground-referenced.
Maximum input voltage: Be aware that the pass device you use must be able to withstand the full raw DC input voltage. Although in practical use the pass device will only see the Vin-Vout differential, this is not the case at the moment you connect a (capacitive) load to the regulator. The load represents a very short term short, until the caps start to charge. This can be enough to destroy a solid state pass device. Tube pass devices will most probably survive this, but you have been warned!