The 10 uF is intended as local supply bypassing on the board. It prevents the regulator IC from oscillating should the end user use very long wires on the input to the regulator.
You'll still need a reservoir cap between the rectifier and the Maida Regulator. Depending on the current draw, you'll need somewhere around 47~220 uF. There's no reason to go higher than that as the regulator will knock the ripple down below the noise floor.
~Tom
You'll still need a reservoir cap between the rectifier and the Maida Regulator. Depending on the current draw, you'll need somewhere around 47~220 uF. There's no reason to go higher than that as the regulator will knock the ripple down below the noise floor.
~Tom
Hi Tom,
Excellent, thanks much! I had totally overlooked D2 when thinking about this circuit.
thanks again,
ben
Built and tested - works great!
Good morning all,
I am currently testing my implementation of Tom's brilliant regulator. My version differs in that it is mostly through-hole components, with the exception of the tantalum capacitor C3A. I also use the TO-220 version of the LT3080 instead of the surface mount as well. I have no fear of surface mount components...just felt like using through-hole for this project.
Below is a photo of it under test. It is being powered by a variable Eico supply at 150 VDC input at 90 VDC output into a switched resistor load, pulling about 40 mA. The MOSFET is mounted to an old computer heatsink.
thanks much,
ben
Good morning all,
I am currently testing my implementation of Tom's brilliant regulator. My version differs in that it is mostly through-hole components, with the exception of the tantalum capacitor C3A. I also use the TO-220 version of the LT3080 instead of the surface mount as well. I have no fear of surface mount components...just felt like using through-hole for this project.
Below is a photo of it under test. It is being powered by a variable Eico supply at 150 VDC input at 90 VDC output into a switched resistor load, pulling about 40 mA. The MOSFET is mounted to an old computer heatsink.
thanks much,
ben
It is much larger than your board - it measures 3.6" long by 1.9" wide. That's the trade-off for going with all thru-hole components, I guess.
Yeah... Mine is 2.50"x2.55". With the thru-hole components you also have to be more careful with the layout if you want to maintain the good performance. Especially the nodes around the Iset current are rather critical. Tradeoffs, tradeoffs.
Still. Nice build. Thanks for sharing.
~Tom
This regulator provides no galvanic isolation, so you will need to use an isolation transformer.
Generating 190 V DC from a 200 V AC supply should be possible, however. You will need to provide a transformer that generates 200 V, a rectifier, and a reservoir capacitor (47 uF is usually enough) before the regulator.
~Tom
Generating 190 V DC from a 200 V AC supply should be possible, however. You will need to provide a transformer that generates 200 V, a rectifier, and a reservoir capacitor (47 uF is usually enough) before the regulator.
~Tom
thanks,
it seems very good. I want to built a prototype but I don't think I can find parts locally
The purpose of D3 is to protect the IC during power-up into a capacitive load and during high di/dt spikes such as those caused by loose connections in tube sockets and the like.
The exact spec on this is not super critical. I would think any unidirectional Tranzorb or TVS type diode with a breakdown voltage around 20 V that fits the DO-214 (SMA) footprint would work. The one you link to looks like a good candidate.
~Tom
The exact spec on this is not super critical. I would think any unidirectional Tranzorb or TVS type diode with a breakdown voltage around 20 V that fits the DO-214 (SMA) footprint would work. The one you link to looks like a good candidate.
~Tom
Hi Tom,
I am also thinking about using your circuit to make an adjustable PS. Some time ago I bought some multiturn 300k/5W pots and these would replace R9.
My worry is about survival chances of the regulator in the case of short circuits. On your website you write "R3 provides a very crude current limiting feature. It is not intended to provide protection against short circuits, but merely serves to help Q1 survive the worst transient currents". That should answer my question, but I am still curious what would happen in the case of a short circuit and if you have any suggestion how to implement a 'better' current limiting circuit?
Many thanks!
Erik
I would probably use a circuit that shorts Vgs on the cascode (Q1) for current limiting. These kinds of circuits can be rather challenging to get stable and they will burn gobs of power in Q1 when the current is being limited.
The 21st Century Maida Regulator will survive an over-current condition, but probably not a straight short circuit. If you can live with "over current protection" rather than "short circuit protection", I'd use the circuit as is.
But if you're designing a lab grade supply, I suggest picking up a couple of service manuals for the HP high-voltage lab supplies and learn their tricks. They use synchronous rectification to limit the power in the pass device as well.
~Tom
The 21st Century Maida Regulator will survive an over-current condition, but probably not a straight short circuit. If you can live with "over current protection" rather than "short circuit protection", I'd use the circuit as is.
But if you're designing a lab grade supply, I suggest picking up a couple of service manuals for the HP high-voltage lab supplies and learn their tricks. They use synchronous rectification to limit the power in the pass device as well.
~Tom
Hi Tom,
thanks for your answer.
In a first attempt of the planned PS I used a tektronix transformer with 5 HV primaries (between 110 and 190V). I connect these in series to get an approximation of the needed voltages. I then have 4x PS, with bridges or doublers and big caps for filtering. The raw DC goes in the regulator, which consists of a LR8N3 feeding the gate of a HV MOSFET. It works, but it was not current protected. I think it wouldn't be hard to implement a current limiter, but then I found your project with the excellent LT3080 and the much better performance (-120dB). It would probably even allow a higher output voltages in comparion to the LR8N3 version, but also no short circuit protection.
I have downloaded the manual of the HP PS. I can grasp the idea of it, but building something like that and getting it stable is not something I will manage, I think.
many thanks! Erik
thanks for your answer.
In a first attempt of the planned PS I used a tektronix transformer with 5 HV primaries (between 110 and 190V). I connect these in series to get an approximation of the needed voltages. I then have 4x PS, with bridges or doublers and big caps for filtering. The raw DC goes in the regulator, which consists of a LR8N3 feeding the gate of a HV MOSFET. It works, but it was not current protected. I think it wouldn't be hard to implement a current limiter, but then I found your project with the excellent LT3080 and the much better performance (-120dB). It would probably even allow a higher output voltages in comparion to the LR8N3 version, but also no short circuit protection.
I have downloaded the manual of the HP PS. I can grasp the idea of it, but building something like that and getting it stable is not something I will manage, I think.
many thanks! Erik
I'm sure someone has already asked this, but just to be on the safe side, is it possible to make a negative version by switching polarity of the parts? (I want to make a +/-130VDC PSU).
Also, since I'm going to build the regulator myself and it's not easy to obtain the parts here, is this possible to use LT1085/LT1033 & IRF630/IRF9630?
Thanks,
Also, since I'm going to build the regulator myself and it's not easy to obtain the parts here, is this possible to use LT1085/LT1033 & IRF630/IRF9630?
Thanks,
To my knowledge, there is no substitute for the LT3080. One of the guys here ran a simulation of two of my 21st Century Maida Regulators for use in a split (±) supply. He reported that the 21st Century Maida Regualtor would only work as a positive regulator.
For ±130 V, a regular Maida Regulator would work OK. You can use LM317 for the positive regulator and LM337 for the negative. You, obviously, won't get the same performance as I get with the 21st Century Maida Regulator, but you'll have a regulated supply. At ±130 V, the power dissipated by the feedback network to ensure 10 mA minimum output current is a lot more manageable.
I can't tell if you can use the IRF630/9630 in your application without running a simulation. It depends on the amount of output current you are designing for. The challenge here is that the device needs to be able to survive start-up. So basically, the MOS device will need to be able to handle Vds = Vin and Id = Iout_max for 100 ms or so. You'll need to consult the SOA curves of the data sheet to figure this out.
There's nothing magic about the STW12NK95Z that I recommend. The only requirement is that this transistor needs to be able to handle the power dissipated in it -- both during start-up (that's the worst case for SOA) and during normal operation (worst case for thermals).
~Tom
For ±130 V, a regular Maida Regulator would work OK. You can use LM317 for the positive regulator and LM337 for the negative. You, obviously, won't get the same performance as I get with the 21st Century Maida Regulator, but you'll have a regulated supply. At ±130 V, the power dissipated by the feedback network to ensure 10 mA minimum output current is a lot more manageable.
I can't tell if you can use the IRF630/9630 in your application without running a simulation. It depends on the amount of output current you are designing for. The challenge here is that the device needs to be able to survive start-up. So basically, the MOS device will need to be able to handle Vds = Vin and Id = Iout_max for 100 ms or so. You'll need to consult the SOA curves of the data sheet to figure this out.
There's nothing magic about the STW12NK95Z that I recommend. The only requirement is that this transistor needs to be able to handle the power dissipated in it -- both during start-up (that's the worst case for SOA) and during normal operation (worst case for thermals).
~Tom
So it looks like I have to go with regular Maida circuits for this bipolar PSU. Anyway I will try breadboard your circuit with LM317/337 to see what will happen in realtime. And if better performance is required, I guess I can always use 2 of your positive circuits (with the price of replacing the transformer to one with separated secondary windings).
About the mosfet, I was concern about the SOA as it is not easy to find a high voltage P-mos here. Luckily I've just found some 2SK310/2SJ117 with better margins so no problem now.
Thanks for your suggestions.
Duong,
About the mosfet, I was concern about the SOA as it is not easy to find a high voltage P-mos here. Luckily I've just found some 2SK310/2SJ117 with better margins so no problem now.
Thanks for your suggestions.
Duong,