nd I'm only using these in one direction, only as a heater.
ver and a FET.
bananas. Basically it looks like the PWM signal from one heater couples on to my rails somehow and then starts to impact the other heaters... I'm not exactly sure if that's what's happening, but they defeintely don't play nic e together, once I plug two in the PWM goes all out of what, bascially beco mes super sensitive and almost floats up to 100% duty cycle when I touch th e potentiometer
ge reference (TLV431)
linear chip)
Folks have mentioned hf spikes getting onto power and ground rails. I might drop this circuit altogether, do the control using hex invertors/opamps/co mparators, and use slowed low fT output transistors to give less switching transient trouble. Slow the whole thing down too, 1kHz heater drive is poin tlessly generating crap.
and I'm only using these in one direction, only as a heater.
river and a FET.
s bananas. Basically it looks like the PWM signal from one heater couples o n to my rails somehow and then starts to impact the other heaters... I'm no t exactly sure if that's what's happening, but they defeintely don't play n ice together, once I plug two in the PWM goes all out of what, bascially be comes super sensitive and almost floats up to 100% duty cycle when I touch the potentiometer
tage reference (TLV431)
he linear chip)
ht drop this circuit altogether, do the control using hex invertors/opamps/ comparators, and use slowed low fT output transistors to give less switchin g transient trouble. Slow the whole thing down too, 1kHz heater drive is po intlessly generating crap.
If you do the PWM and control digitally, the spikes are less of problem - c omparators and spikes aren't a good mix. It takes much bigger spikes to scr ew up counters and logic than it does to screw up comparators.
Chopping at 1kHz can make your filtering problems a bit difficult - we ende d up doing the PWM at 17kHz which which still left the inductor a bit bulky . A proper sigma-delta modulation scheme can - in theory - push most of the switching noise up to frequencies which are relatively easy to filter out, but it's trickier to be sure that your switching transistors won't get too hot.
ine, and I'm only using these in one direction, only as a heater.
ET driver and a FET.
goes bananas. Basically it looks like the PWM signal from one heater coupl es on to my rails somehow and then starts to impact the other heaters... I' m not exactly sure if that's what's happening, but they defeintely don't pl ay nice together, once I plug two in the PWM goes all out of what, basciall y becomes super sensitive and almost floats up to 100% duty cycle when I to uch the potentiometer
voltage reference (TLV431)
on the linear chip)
might drop this circuit altogether, do the control using hex invertors/opa mps/comparators, and use slowed low fT output transistors to give less swit ching transient trouble. Slow the whole thing down too, 1kHz heater drive i s pointlessly generating crap.
comparators and spikes aren't a good mix. It takes much bigger spikes to s crew up counters and logic than it does to screw up comparators.
ded up doing the PWM at 17kHz which which still left the inductor a bit bul ky. A proper sigma-delta modulation scheme can - in theory - push most of t he switching noise up to frequencies which are relatively easy to filter ou t, but it's trickier to be sure that your switching transistors won't get t oo hot.
There's no need to solve a problem if you dont create it. For 60Hz or less switching, why switch hard & fast.
fine, and I'm only using these in one direction, only as a heater.
FET driver and a FET.
it goes bananas. Basically it looks like the PWM signal from one heater cou ples on to my rails somehow and then starts to impact the other heaters... I'm not exactly sure if that's what's happening, but they definitely don't play nice together, once I plug two in the PWM goes all out of what, bascia lly becomes super sensitive and almost floats up to 100% duty cycle when I touch the potentiometer
a voltage reference (TLV431)
n on the linear chip)
I might drop this circuit altogether, do the control using hex invertors/o pamps/comparators, and use slowed low fT output transistors to give less sw itching transient trouble. Slow the whole thing down too, 1kHz heater drive is pointlessly generating crap.
- comparators and spikes aren't a good mix. It takes much bigger spikes to screw up counters and logic than it does to screw up comparators.
ended up doing the PWM at 17kHz which which still left the inductor a bit b ulky. A proper sigma-delta modulation scheme can - in theory - push most of the switching noise up to frequencies which are relatively easy to filter out, but it's trickier to be sure that your switching transistors won't get too hot.
s switching, why switch hard & fast.
You switch rapidly because it minimises the dissipation in the switching de vice. The shorter time that there's a current flowing through the switching device while there's a voltage drop across the switching device, the lower the energy dumped in the switching device during switching.
You switch frequently, because the capacitors and inductors required to div ert and block the high frequency content of the PWM current become smaller and cheaper as the frequencies being blocked become higher. Particularly wh en driving Peltier junctions, you don't wanted to put the raw switched curr ent straight through the Peltier junction. The heat transferred by the junc tion is a linear function of the current put through it, but the ohmic heat ing within the junction is proportional to the square of the instantaneous current.
Having high frequency currents circulating through the - usually long - lea ds to the Peltier device isn't a good idea either. They radiate nosie to t he rest of the circuit, and the rest of the world.
Den tirsdag den 12. august 2014 01.59.02 UTC+2 skrev Maynard A. Philbrook J r.:
:
moelectric coolers (peltier cooler)
at's fine, and I'm only using these in one direction, only as a heater.
side FET driver and a FET.
... it goes bananas. Basically it looks like the PWM signal from one heater couples on to my rails somehow and then starts to impact the other heaters ... I'm not exactly sure if that's what's happening, but they definitely do n't play nice together, once I plug two in the PWM goes all out of what, ba scially becomes super sensitive and almost floats up to 100% duty cycle whe n I touch the potentiometer
ut
from a voltage reference (TLV431)
ption on the linear chip)
ils. I might drop this circuit altogether, do the control using hex inverto rs/opamps/comparators, and use slowed low fT output transistors to give les s switching transient trouble. Slow the whole thing down too, 1kHz heater d rive is pointlessly generating crap.
blem - comparators and spikes aren't a good mix. It takes much bigger spike s to screw up counters and logic than it does to screw up comparators.
we ended up doing the PWM at 17kHz which which still left the inductor a b it bulky. A proper sigma-delta modulation scheme can - in theory - push mos t of the switching noise up to frequencies which are relatively easy to fil ter out, but it's trickier to be sure that your switching transistors won't get too hot.
less switching, why switch hard & fast.
g device. The shorter time that there's a current flowing through the switc hing device while there's a voltage drop across the switching device, the l ower the energy dumped in the switching device during switching.
what was wrong with it?
you want to switch hard and fast because you are dissipating energy in the switching device during the transition
though if your switching frequency is low it may not be a big issue because it doesn't happen very often
On Tuesday, 12 August 2014 09:59:02 UTC+10, Maynard A. Philbrook Jr. wrote :
:
less switching, why switch hard & fast.
g device. The shorter time that there's a current flowing through the switc hing device while there's a voltage drop across the switching device, the l ower the energy dumped in the switching device during switching.
Please get somebody who does complex sentence comprehension to read what I wrote, and ask them to explain it to you. That paragraph was totally non-co ntroversial.
I once rewrote the section on photon detection, but I was never interested enough to go back and see if my edits had stuck.
divert and block the high frequency content of the PWM current become smal ler and cheaper as the frequencies being blocked become higher. Particularl y when driving Peltier junctions, you don't wanted to put the raw switched current straight through the Peltier junction. The heat transferred by the junction is a linear function of the current put through it, but the ohmic heating within the junction is proportional to the square of the instantane ous current.
leads to the Peltier device isn't a good idea either. They radiate noise to the rest of the circuit, and the rest of the world.
Finding your wits might be a more useful exercise. You do seem to have give n at least half of them away. Here's a message that might be short and clea r enough for you to comprehend.
"Rapidly" can have two different interpretationshere : "fast rise/fall times", or "frequently".
The former is good, as far as reducing dissipation goes... faster rise and fall means less time per cycle in the transition zone, where the dissipation is higher.
The latter is bad, as far as reducing dissipation goes... more transitions per second means a higher fraction of total time is spent in the transition zone (for any given rise and fall time).
So, using a phrase like "switch rapidly" is probably not the best of ideas... especially when posting in a forum where there's almost certainly going to be someone who will leap upon the slightest ambiguity in any poster's words, and go right for the throat and entrails ;-)
switching device. The shorter time that there's a current flowing through the switching device while there's a voltage drop across the switching device, the lower the energy dumped in the switching device during switching.
he switching device during the transition though if your switching frequenc y is low it may not be a big issue because it doesn't happen very often
mes", or "frequently".
Which I disambiguated in my post - the one that excited Jamie's uncomprehen ding response - by going on to talk about why you switched frequently.
d fall means less time per cycle in the transition zone, where the dissipat ion is higher.
s per second means a higher fraction of total time is spent in the transiti on zone (for any given rise and fall time).
s... especially when posting in a forum where there's almost certainly goi ng to be someone who will leap upon the slightest ambiguity in any poster' s words, and go right for the throat and entrails ;-)
The problem here wasn't the ambiguity in what I posted, but Jamie's incapac ity to make sense of it. He's got a problem with processing complex sentenc es and would like it if everything that was posted here came up as the simp le declarative sentences that he can - more or less - follow.
Sadly, he doesn't ask for grammatically less demanding expositions, but mer ely complains that he can't follow information that isn't packaged as acces sibly as he'd like. I doubt that he understands what he is complaining abou t - his subjective sensation of incomprehension seems to form the core of h is complaint, but he isn't articulate enough to make this explicit.
For some that may not actually work hand and hand they may not see anything wrong with his explanation, for me, it looks look a data collector's SE edition (Shoulder Educated).
I noticed Davide Platt responded with the exact image of the wording as I got and as far as I am concern, slowman believes everything he spits out, no matter where he collects his info.
On Tuesday, 12 August 2014 11:39:26 UTC+10, Maynard A. Philbrook Jr. wrote :
ching device. The shorter time that there's a current flowing through the s witching device while there's a voltage drop across the switching device, t he lower the energy dumped in the switching device during switching.
the switching device during the transition though if your switching frequen cy is low it may not be a big issue because it doesn't happen very often
An unsubstantiated claim.
ing wrong with his explanation, for me, it looks look a data collector's SE edition (Shoulder Educated).
In other words Jamie can't understand it, and can't specify what's wrong wi th it that makes it difficult for him to comprehend.
I got and as far as I am concerned, Sloman believes everything he spits ou t, no matter where he collects his info.
Most of it came from
Sloman A.W., Buggs P., Molloy J., and Stewart D. "A microcontroller-based d river to stabilise the temperature of an optical stage to 1mK in the range
4C to 38C, using a Peltier heat pump and a thermistor sensor" Measurement S cience and Technology, 7 1653-64 (1996)
which I wrote clearly enough to get it published. Scholar.google lists it a s cited by 16, only two of whom are me. Jamie claims to think that I believ e everything that I post - which is a little silly of him, since I do post the occasional exaggeration for comic effect, and the even rarer occasional mistake - but Jamie's thinking isn't all that reliable.
device. The shorter time that there's a current flowing through the switchi ng device while there's a voltage drop across the switching device, the low er the energy dumped in the switching device during switching.
Of course, but that doesn't mean you therefore need hard fast switching. If the heating device is run off the transformer secondary or mains, ie on ac , you can switch near zero crossings without creating spikes or significant P_diss in the switch.
ivert and block the high frequency content of the PWM current become smalle r and cheaper as the frequencies being blocked become higher. Particularly when driving Peltier junctions, you don't wanted to put the raw switched cu rrent straight through the Peltier junction. The heat transferred by the ju nction is a linear function of the current put through it, but the ohmic he ating within the junction is proportional to the square of the instantaneou s current.
Now I know a bit more about peltiers.
NT
eads to the Peltier device isn't a good idea either. They radiate nosie to the rest of the circuit, and the rest of the world.
m - comparators and spikes aren't a good mix. It takes much bigger spikes t o screw up counters and logic than it does to screw up comparators.
ended up doing the PWM at 17kHz which which still left the inductor a bit bulky. A proper sigma-delta modulation scheme can - in theory - push most o f the switching noise up to frequencies which are relatively easy to filter out, but it's trickier to be sure that your switching transistors won't ge t too hot.
less switching, why switch hard & fast.
less useful work with the Peltier. Higher frequencies are cheaper to filter out.
"Ripple on your DC" isn't usually big enough to generate enough extra ohmic heating to be worth worrying about. "Bang-bang" control of the current thr ough your Peltier junction - from maximum to nothing - is what usually does the damage, and - for instance - generated the story my boss told me in 19
93 about melting the solder inside a Peltier assembly, so that the top laye r of alumina went one way and the bottom layer another. An expensive mistak e, not to mention obvious and embarrassing.
The seductive idea of picking when you turn the current through the Peltier on or off becomes less seductive when you realise that you have to keep th e current through the junction pretty much stable if you want the Peltier j unction to shift any significant amount of heat.
And "ripple on your DC" which is too small to generate any extra ohmic heat ing worth worrying about can still be high enough to radiate inconvenient a mounts of EMI into sensitive bits of adjacent circuitry.
On Wednesday, 13 August 2014 08:55:33 UTC+10, Maynard A. Philbrook Jr. wro te:
ed driver to stabilise the temperature of an optical stage to 1mK in the ra nge 4C to 38C, using a Peltier heat pump and a thermistor sensor" Measureme nt Science and Technology, 7 1653-64 (1996)
it as cited by 16, only two of whom are me. Jamie claims to think that I be lieve everything that I post - which is a little silly of him, since I do p ost the occasional exaggeration for comic effect, and the even rarer occasi onal mistake - but Jamie's thinking isn't all that reliable.
does not vindicate you.
I don't need vindication. Your inability to see the relevance of what I pos ted makes your cognitive inadequacies embarrassingly obvious.
Not a useful activity, but at least it keeps you out of our hair.
blem - comparators and spikes aren't a good mix. It takes much bigger spike s to screw up counters and logic than it does to screw up comparators.
we ended up doing the PWM at 17kHz which which still left the inductor a b it bulky. A proper sigma-delta modulation scheme can - in theory - push mos t of the switching noise up to frequencies which are relatively easy to fil ter out, but it's trickier to be sure that your switching transistors won't get too hot.
or less switching, why switch hard & fast.
do less useful work with the Peltier. Higher frequencies are cheaper to fil ter out.
hmic heating to be worth worrying about. "Bang-bang" control of the current through your Peltier junction - from maximum to nothing - is what usually does the damage, and - for instance - generated the story my boss told me i n 1993 about melting the solder inside a Peltier assembly, so that the top layer of alumina went one way and the bottom layer another. An expensive m istake, not to mention obvious and embarrassing.
of the current squared.
Size does matter. A little bit of extra heat is one thing. Enough extra hea t to let the Peltier array dismantle itself is a different ball-game.
nd I'm only using these in one direction, only as a heater.
ver and a FET.
bananas. Basically it looks like the PWM signal from one heater couples on to my rails somehow and then starts to impact the other heaters... I'm not exactly sure if that's what's happening, but they defeintely don't play nic e together, once I plug two in the PWM goes all out of what, bascially beco mes super sensitive and almost floats up to 100% duty cycle when I touch th e potentiometer
ge reference (TLV431)
linear chip)
Much thanks for all the replies! I've cleaned it up enough to where I'm hap py with it, it's not ideal but good enough for my application.
I think my main problem was
- The MOSFET Driver IC, I think it was overkill for what I'm doing and it w as pulling on my supply voltage, especially when trying to run several heat ers (TECs)
Solution:
- I replaced the MOSFET driver IC with a MOSFET (2 actually to invert my si gnal, but 1 solved it) and I added 2,000uF of capacitance
- Added a cap across the FET, that helped the ringing from the TEC
- And finally, added 2,000uF on my 12V bus, without this every time a TEC s witched on it would pull my 12V down, sometimes significantly depending on the overlap of the duty cycles and the duty cycles
Well, it'll take some 5-10us to turn on, so that'll also solve part of your problem. Turn-off is still fairly fast though (~100ns?). Why would you skip a gate driver proper, when you can just increase the gate resistor to 100 or 1k?
Sure, if you want it to start a fire. Tantalum besides? You neither need nor want that there.
If you want to dampen ringing, use the smallest C possible, with a series resistor. 10 ohm + 0.01u would be a good starting point.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.