It's virtually impossible for a regulator not to behave inductively as the frequency goes up. Yes, one has to be careful with the output capacitor. I would not go as far as to say, "cannot be use."
All of them are. It depends how fast, as a function of frequency, the impedance goes up. That will be how they differ.
As far as v2 goes, I recommended as small a value as possible for the output capacitor. It's difficult to predict how low the ESR should be. Everyone may build the regulator differently, some p2p, others with their own pcb, resulting in different inductive behaviour for each and hard to predict the interaction with the cap ESR.
In my case the output capacitor is not intended to do this. In v2 it is for stability.
Look at the plot of gain vs. frequency. There are no miracles.
This is true of any regulator, no matter whether it's bypassed with a cap or not.
I think the real question is how much we can get away with, in a real circuit.
There is the simplistic HV, but that one is a different, not strictly V1 style, whimsical circuit, mostly anecdotally documented. People like it very much for some reason, but its not technically that serious or reliable, beyond what is tested to work with till now, by some members later and me initially.
Here are the pictures.
I prefer using individual heatsink for individual MOSFET as a habbit, as I am worried about the high capacitance between the MOSFET and heatsink (think about how a capacitor is made from). Sharing a heatsink between MOSFETs means connecting them with a capacitor. The MOSFET used here are large TO247 type. Fortunately, the drains would be connected so there would be no issue. This is not the same with v2. Nevertheless, I am lazy. I don't want to do extra drilling so I got these cheap screw-on heatsinks. They don't look like 6 degree / C thermal rating they have on the badge.
I also add in the thermal pad to isolate the drain from the heatsink, not necessary but in some cases may be helpful.
Provided that the thermal pad adds 2 degree / C making it 8 degree C, consider that I will use around 350mA on 15VDC rails the power disappation may be up to 6W per MOSFET. So the maximum temporature would be up to 50 degree higher than room temporature, let us say 80 degree at maximum. Since the MOSFET can operate up to 150 degree (bad performance though), I guess it would be fine. In real life, the figure would be a bit lower.
I prefer using individual heatsink for individual MOSFET as a habbit, as I am worried about the high capacitance between the MOSFET and heatsink (think about how a capacitor is made from). Sharing a heatsink between MOSFETs means connecting them with a capacitor. The MOSFET used here are large TO247 type. Fortunately, the drains would be connected so there would be no issue. This is not the same with v2. Nevertheless, I am lazy. I don't want to do extra drilling so I got these cheap screw-on heatsinks. They don't look like 6 degree / C thermal rating they have on the badge.
I also add in the thermal pad to isolate the drain from the heatsink, not necessary but in some cases may be helpful.
Provided that the thermal pad adds 2 degree / C making it 8 degree C, consider that I will use around 350mA on 15VDC rails the power disappation may be up to 6W per MOSFET. So the maximum temporature would be up to 50 degree higher than room temporature, let us say 80 degree at maximum. Since the MOSFET can operate up to 150 degree (bad performance though), I guess it would be fine. In real life, the figure would be a bit lower.
An externally hosted image should be here but it was not working when we last tested it.
Top view:
Comparing to the PCB design:
An externally hosted image should be here but it was not working when we last tested it.
Comparing to the PCB design:
An externally hosted image should be here but it was not working when we last tested it.
Bottom View:
An externally hosted image should be here but it was not working when we last tested it.
Today I replaced the trimpot with 220R (+) and 180R (-). The stablized voltages are now +13.07VDC and -13.17VDC. I can't be more happier with that. Of course, when the new batch of 2SK170 arrives I will replace the njfet.
It still took some time to stablize the negative voltage so the problem is not the trimpot.
Finally, a picture showing my latest work: I removed the "input" and output capacitors from the board. Solder some wires on their positions, and solder the capacitors on the stand-out wires. This got it ready to do capacitor tests, as it is much easier to solder and desolder them without disconnecting the boards. They may look like antenas, but they sound OK.
I have already got some very interesting test results. But I won't post them until I finish all the tests.
It still took some time to stablize the negative voltage so the problem is not the trimpot.
Finally, a picture showing my latest work: I removed the "input" and output capacitors from the board. Solder some wires on their positions, and solder the capacitors on the stand-out wires. This got it ready to do capacitor tests, as it is much easier to solder and desolder them without disconnecting the boards. They may look like antenas, but they sound OK.
I have already got some very interesting test results. But I won't post them until I finish all the tests.
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
Very orderly work, good base for clean experiments. When you will change the NJFET see to find one with about the same IDSS. Some difference leads to another Vset resistor value for same Vout of course. If it will still take some time to settle, means it is something (Tc and time related most probably, not terrible) with the negative side. Not in -Vref only by definition. Is it about 0.6V the total slide? That BC560C is nearer than the 550C, talking buffer part. Could be more Tc susceptible for instance. No big deal in my book anyway, if it is not associated with some instabillity. Without instruments, you would hear periodic whistle and buzz if it was something serious. You did well to isolate, once, early in the phono thread someone pushed together the individual sinks, got a zap and a bang, but the IRFPs did not blow. Tough fellows. Also for getting a bit hot, they aren't the ones who gonna riot sooner than others. Now you tune your sound, good part of the job. It is nice that someone is going to listen a lot to versions in a big system, tune 'em, recommend uses etc. Good info. Cheers.
Checked it.
I'm evaluating shunt regulators circuits for a low-wattage class A amplifier dedicated to HF. The voltage requirement is 17V, amperage around 1A.
Which values would you suggest for my needs?
Thanks for the input, and Salas too.
I think I will build a BJT version first as I have some BD140s on hand, then convert it to mosfet at a later date.
If you use 1N4738A for D1 and D2 you get around 16.7V output. You need to adjust R2 to get the current up to the value you want. That needs to be done experimentally. You can use a 10K trimmer for R2, and start at about 2K. If you're running 500uA through R2 then somewhere between 8k and 10k will give you the 1.xA current limit that you want.
It's not exactly a beginner's project, you need to be comfortable with applying Ohm's law left, right, and center.
(I'm not kidding. 
)
It'd be best to choose a target schematic, either v1, or v1.1, or v2. I'd recommend v1.1 or v1, to get your feet wet. Then choose your target voltage and current. Then, if you really want to use the BD140, someone needs to make sure it works in that schematic.
Sure, I'm looking to build something close to V1, with up to 100mA total current. Load current is around 40mA. At worst it may oscillate, at which point I'll have to do something to decrease the gain of the circuit.
The trouble with getting a logic level mosfet is getting it in low quantities without paying ridiculous postage. I'll order some next time I need a batch of components from Farnell. Famous last words, but I can't see why the BD140 wouldn't work. There are plenty of shunt schematics out there, many of which use a PNP in a similar configuration. I'll report back.
The trouble with getting a logic level mosfet is getting it in low quantities without paying ridiculous postage. I'll order some next time I need a batch of components from Farnell. Famous last words, but I can't see why the BD140 wouldn't work. There are plenty of shunt schematics out there, many of which use a PNP in a similar configuration. I'll report back.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.