Pololu Robotics & Electronics
My account Comments or questions? About Pololu Contact Ordering information Distributors

Pololu Forum

24V Shunt Regulator not working as expected

Dear all,
I’ve bought two Shunt Regulator: 26.4V, 2.80Ω, 15W https://www.pololu.com/product/3775/
but they don’t work as expected.

If the VIN Voltage rises above 26.4V, the Gate-Source Voltage of the MOSFET (AON6232) is only 2,4V which results in VDS approx. 24V so the MOSFET has to dissipate 90% of the power (and thus getting hot really quickly) and the shunt resistors only 10%. I really doubt this was intended by the designer.

Btw, can someone tell me what device is the one labeled wit “GC9” I guess a voltage detector with 2.5V threshold (something like ROHM BD48L25) but I can’t find a device with SOT-23 package which would match the used connection.

– Andy

Hello, Andy.

We do not disclose or commit to the components on that product. In what sense do the boards not work as expected? (Is there an external problem or do the internal workings of the circuit just not match what you expected?)


Hi Nathan,
as I wrote above the MOSFET does get hot very fast (and I described why this is the case). Two shunt regulators are already damaged while decelerating a 40W stepper motor (kinetic energy approx. 100J)

I understand that you want to “protect” your product when not opening the schematic but for me as engineer it would help to find the problem. And thus I’m really disappointed that you don’t even want to disclose one single device.

Nevertheless, the summary is: Two shunt regulators are now damaged because what I think is a big design flaw


I deleted your edit in which you added extra information about your speculation about the parts used. Before moving on to technical details, I want to comment on your behavior in case you do not realize how poorly it is coming across to us.

You are a guest on this forum. We made it clear we do not want to give out the details of the parts involved, and you proceeded to post them (or something close to that) here. I really do not understand what you thought you were going to achieve by doing that. If I went to your house and you asked me not to do something there, I would not do it or I would leave, independent of my assessment of how reasonable your request was.

Regarding your broader response, we have no problem with criticism and your expressing your perspective even if it’s negative or disappointed or whatever. Obviously, we are aware that people prefer us disclosing all kinds of things. Likewise, you must be aware of at least some reasons to keep some information private and that there might be additional reasons you are not aware of. If there were some obvious right answer for how to weigh various pros and cons, everyone would probably be doing the same thing. It’s helpful for you to add weight to a pro or con (even if it’s as basic as saying, hey, I would really appreciate it if you guys shared more details); focusing on just one side as if it’s the only factor is not very useful.

On to the more technical aspects of your posts: Nathan had a much longer initial response with all kinds of details, and I reminded him that he is not supposed to be putting a lot of effort into a particular assumption without getting some basic information. “The MOSFET does get hot very fast” does not give us much to work with, especially for an item that is expected to get hot. How hot? How fast? And the thing is, we’re not necessarily asking for that (though it could be helpful). But “the MOSFET went up in flames”, or even the less dramatic, “the MOSFET broke”, at least identifies a problem.

I appreciate that you included the specificity of the 100J, which might be a clue that that is just too much for this product (for instance, if that 100J has to get dumped over 5 seconds, that would be 20W, which is beyond the limit of this regulator). Your guess about the component involved (which I deleted) should give you a hint that the regulator probably does not work exactly as you initially assumed (the MOSFET is intentionally getting used in linear mode), so if there is a problem with the design, it’s not exactly what you suggested in your posts.

So, it’s possible these regulators just will not have the power capacity your application needs, in which case there is not that much we can do. It’s possible the external resistor version could work. If you want to troubleshoot this more, please confirm that the MOSFETs indeed went up in smoke and provide some information about your system.

- Jan

Dear Jan, thank you for your answer.
I tried to start a technical discussion in my first post but I’ve interpreted Nathans answer as “we don’t want to help” and as an attempt to feign ignorance .

You can’t tell me that the MOSFET it intentionally used in linear mode at the intended operating point. For your product 3775 there is no description in the specs how long it can dissipate 15W. So please go ahead, connect it to a power supply, limit current to 0.5A and increase voltage above 26.4V you’ll see the MOSFET damaged in less than 10s.

All I got from the discussion is: “We don’t want to help or discuss problems with our products”
Perhaps this wouldn’t be my impression if you haven’t hold back Nathan in his first reply.

For me this is end of discussion here because I don’t think this can constructive. Feel free to delete the entire posting, all I can do is to tell others from my bad experience.

– Andy

This is from the 3775 product page, seems relevant to this topic:

Hi Jilo, thanks and you are right, It IS on the description page, not on the “Specs” tab

Your input has been helpful in pointing out the problems associated with evidently misunderstanding the purpose of the product, protection against very short term voltage spikes, and the consequences of abusing it by applying longer term overvoltages.

Your ill advised proposal to test the device demonstrates this point:

connect it to a power supply, limit current to 0.5A and increase voltage above 26.4V you’ll see the MOSFET damaged in less than 10s.

I think this thread is useful in various ways. We probably identified the root problem, which others might also have in the future, and I am happy to be judged by our response here, including revising our support guidelines based on any feedback we get. Andy, if you must share your experience as bad, please direct people to this thread so they can judge for themselves.

For anyone looking at this thread, I would like to make clear that the only edit I made to Andy’s posts is to remove one out-of-context (and in my view, gratuitous) line that said something like, “looks like the part is <partnumber>”.

- Jan