What is the best heat transfer method

[richardb]

ChrisP58 I was pointing out a feature of the hip4081 in that you don't need to drive the high side inputs as there is electronics in the chip that cope with shoothrough.

ronv you should have had a regulated supply for the chip I certainly run well above 40V on the fets with no problem and at 200kHz and the fets are quite large.
The psu is 400W with no heatsink on the FETs

 

[ronv]

Hi rich,
Yes it was my problem. I was using the same supply (16v @ 6 amps) for both the motors and the HIP. It was regulated but nothing except capacitors to keep the voltage from going to high when the motor acted like a generator. Not a problem when on batteries.
Jeff, Putting the FETs a long way from the driver is not a good idea.

 

[kinarfi]

I debated with my self as to whether or not to make this post, I have redesigned, rebuilt and about to test my power steering controller and FET driver, I took so photos, but don't want to burden ETO with them yet, so I put them in dropbox for now, If anyone sees any possible problem, please let me know, I do plan to shorten the wires from circuit board to plug and plug to FETs. One app sheet I read suggested adding a ferrite beads to reduce parasitic oscillations, so I did and another on said to add a cap from drain to gate(I think) but I can't it right now, so I'm not doing that.
https://www.dropbox.com/sh/803pr9babscxq6m/JQYpryqada

 

[kinarfi]

Today, I tried to use my HIP4081 and something is not right, I bought 2 and one may already be dead, not sure, on #2, I applied 10v to pin 3, dis, and measured the voltage on all pins, starting with 1 I measured:13.00, 13.26, 10, 0, 0.2, 0.2, 13.24, 5.04, 5.04, 13.00 #11 - #14, 10.07 15&16, 13.24,#17-#20 10.07. I thought all output were supposed to go to 0.
I then removed the 10v to pin 3 and my measurements were:13.25, 13.26, .584, 0, .201, .201, 13.26, 5.03, 5.03, 13.25, #11 12.84, 11.39, 11.80, 11.81, 13.26, 13.26, 11.81, 11.80, 11.39, 12.83.
I then applied a PWM signal to pin 5, then pin 6 and looked at all outputs with an oscilloscope and all outputs had the same out put, it matched the input, but much less amplitude than the input and it was above the input. NOTE, when ever I applied power to the fets, they seemed to be shorting, but when I removed the driver from the circuit, no shorting, so all measurement were with no power applied to the FETs.
Question, do I need a gate to source resistors, how do I determine the value of the resistor attached to pins 8 & 9?,,The drawing shows a signal going to pin 6 and an inverted signal going to 5, does it need the inverted signal?
thanks for your help,
Jeff

 

[richardb]

I don't really have time to reply properly re the hip output signals as I find it hard to follow without I diagram. but have you done the same test without the chip in place?


1. check the wiring.

2. check the wiring again.

3. I agree with ronv long wires are a bad idea for driving fets but if you have to then use twisted pairs and put the gate drive resistor at the driver end not the fet end

4. whats the gate to source resistor for. ive never fitted one as the output impedance of the fet driver is very low so it will have no significant effect.

5. there is a chart on the datasheet but your running at low frequencies you may as well just run with 250k res which is the max delay shown.

6 . could do with a diagram from the hip through to the motor, even if its a photo of a paper drawing.

7. the cct can be used in 2 ways with motors,

half bride 1 is inverted wrt half bridge 2so on a 100v bus with 10% pwm on bridge 1 the average voltage is 10v on bridge 2 the average voltage is 90v and at 90% the reverse is true.
or
bridge1 is used to vary the voltage and bridge 2 is used to reverse the motor plarity.

8 whats the big bridge rectifier for?

9 make sure all your diodes are fast diodes even the ones on the hip4081

 

[kinarfi]

AAAAAAHHH
Here is a drawing and photos of what I have, I will take measurement again later with and without the chip.
The reason I think I need the Gate to Source resistor is that while I was working on one part of the circuit, I had the FETs sitting with power applied and notice the supply was starting to show current flow. One other consideration is that the FETs will have power applied all the time, but the rest of the circuit will only have power applied when driving.

cct ?
bridge rectifier is the diode across each FET to protect against inductive kick back.

 

[ronv]

It may be toast. But having said that. Als and Bls should not be at 10 volts since they should be tied to ground at the source of the lower transistors.
A couple of observations:
See fig. 32 for the power on reset. Or better yet use the 4080A
The long wires from the HIP board to the FETs probably won't work. I would build the HIP on the same board as the FETs and have the wires between the comparator and the input to the HIP.
Add a little gate resistance - maybe 12 ohms.
When things are working you won't need gate to source resistors. Maybe while you are debugging 250k so they will be off with the HIP disconnected.
Maybe put a resistor where the motor normally goes or a while.

 

[ronv]

I don't know if you saw my edit. The HIP4080A has built in power on reset. Maybe you actually ot that one. (I hope so)
https://www.intersil.com/content/dam/Intersil/documents/an94/an9404.pdf

 

[kinarfi]

What about the inputs, what I have is two separate signal that vary from 0 to 100% duty cycle. Both signal start at 0% , one goes up in duty cycle % and that drives my motor in one direction which pulls the signal back to 0% then the other can do the same or the first can repeat, but the signals always go back to 0% and must pass through 0% to reverse, and one of them is always at 0%

I did shorten my wires, BUT, I had to think outside the box, THAT's a PUN
thinking out side the box.
Still not working, but you may have pointed out my fault, using 2 power supplies, one to act as the KEY and one to act as the battery, I haven't tied the ground/negatives together, which could not happen with a vehicle. I checked the voltages again and they were the same +/- , and with out the chip , the 13s stayed 13+, inputs 5&6 stayed .202,4 stayed 0, 8, 9, 11 - 14 & 17-20 were 0.

 

[ronv]

That's much better. At least everything is off instead of everything being on. Lot less smoke.
The output should follow the inputs. In other words if BHI is high BOH should be high. Since AIH BIH are tied to +12 both AOH & BOH should be at the power supply voltage + about 10 or 12 volts.
Assuming you have a HIP4080A and not just a HIP4080. A couple of possibilities:
1-You have a HIP4080.
2- It may not like the 10K gate to source on the upper transistors. (The bottoms don't care)
3- Disable is low.
4- Something else.
It may work with those long wires, but there are a lot of warnings to keep them really short - like 1/2 inch.
So check disable one more time and remove the 10Ks on the top transistors. This is safe since even if they turn on with everything unplugged the bottoms should be off.
PS. what's the voltage on pin 1 now?

 

[kinarfi]

Definitely a HIP4081A

 

[ronv]

Good good. I attached the wrong app note above. Here is the correct one.
https://www.intersil.com/content/dam/Intersil/documents/an94/an9405.pdf
So did you try it without the 10K gate to source on the uppers?
Might also add 220k resistors for HDEL and LDL.

 

[kinarfi]

WWWWAAAAAA! 2 DOWN, 0 TO GO
I got a couple FAN7842, guess I'll try them.
Mike said invert the signal to the second. I think I see a problem coming when I reverse the direction. I sim and see what I see.

 

[ronv]

The trouble with the FAN is it doesn't have a charge pump, just a bootstrap capacitor to get the top FET voltage higher than the supply. So it won't work at 100% duty cycle.
I can't remember why didn't we use PFETs for the uppers and forget about all this other stuff?
Could you post your sim or a picture of the inputs to the driver? What happened to the last one?

 

[kinarfi]

I have lots of PFETs and I actually had Pfets mounted. What's the best way to drive them? The nice thing about the HIP is that I only needed 1 input per rotation direction. Also, I doubt if I ever really get to 100% duty cycle. Listed among the dead TODAY is 1 555, 1 IRF3205, & 1 HIP4081A.

Question about FAN and IR2104, the IR has a SDnot and the FAN has Lin, Is there any difference, What does Lin do?
Spice has a model for IR2104 that I'm using as My FAN

 

[kinarfi]

Plans for tomorrow, clean up this drawing, build this drawing, see if it works.

 

[richardb]

I could see a bridge rectifier on the heatsink but its not on your diagram what does it do ?
It looks like a standard mains frequency full bridge, if it is don't use it any ware in the high frequency cct(circuit) the diodes take about a week to turn off well many microseconds. don't forget the mosfets have internal diodes you don't need anything extra. but you do need a close couples cap across the rail you also need one across the hip4081 which isn't on your diagram. I would put a low-ish value res in series with the supply and the chip plus a zener across the ic maybe 100R and a 12V zener and a 1uF cap.

also your diagram shows the delay inputs tied to ground !

 

[kinarfi]

I was using the bridge as free wheeling diodes -- it's OUT -- Thanks for the advice!!
I have killed 2 HIP4081a and don't have or plan on getting any more.

I simmed a IR2104 (because it's the only model I could find) as a FAN7842 and if they both work the same way, I should be good to go. The only difference I see is Pin3, the IR2104 has SD-not and the FAN7842 has LIN
The sim runs beautifully and does just what I want,

 

[ronv]

Something looks funny. The HO is going to the lower transistors and the top ones are PFETs. I suppose it's possible that works, but I think the gate voltage for the bottoms might be close to 24 volts?
What is the FB signal in the sim? Feedback from somewhere?

 

[kinarfi]

I think it might have worked, the cap was to small to develop much of a charge and it did sim well, I just didn't work well and I gave up on drivers for the time being, I put a totem together and it ran a P & an N FET nicely, so, I'll try it out tomorrow, nice & simple.
FB is the feed back from a LEM hx sp-2 current transducer, starts at 2.50v and goes up or down at 0.0125 volts per amp, instantaneously so it follows the 30KHz frequency and PWM%. Not sure just what I plan to do with it this time around.
What's you favorite totem set up?
I simmed a couple and spice doesn't seem to like them. I have never had a good sim on a totem.