2N3904 vs 2N3906 Rise and Fall times

Thread Starter

Tonyr1084

Joined Sep 24, 2015
7,853
Looking at the data sheets for the 2N3904 and 3906 transistors.

What does "Delay" "Rise" "Storage" and "Fall" times mean? Aside from the obvious - they have to do with timing. Most curious about "Storage". (see below for my issue)

_3904 ________________ 3906_
-35 nS ---- Delay --- 35 nS
-35 nS ---- Rise ---- 35 nS
200 nS -- Storage - 225 nS
-50 nS ---- Fall ----- 75 nS

This circuit (in part) (below) is operating at 12 volts. R2 is a square wave input (high / low). The 3906 is getting warm. After a minute of operation I'm measuring a case temperature of 99˚F. This is concerning! The red line is 12 VDC (no resistance) and the black line out of the bottom of the 3906 goes direct to ground. This should only be charging and discharging an IGBT Gate.

The IGBT gate capacitance is:
Cies --- Input Capacitance --------------- 5770 -- pF VGE = 0V /// VCC = 30V
Coes -- Output Capacitance ------------- 400 --- pF
Cres -- Reverse Transfer Capacitance ---- 100 --- pF

1579639922763.pngCut and pasted from a larger diagram. Available if necessary - but many have already seen it.
 
Last edited:

Chris65536

Joined Nov 11, 2019
270
I'm not sure about why the transistor is getting hot, but shouldn't your PNP and NPN be swapped? How can there be any emitter current with your design?
 

Thread Starter

Tonyr1084

Joined Sep 24, 2015
7,853
I'm not sure about why the transistor is getting hot, but shouldn't your PNP and NPN be swapped? How can there be any emitter current with your design?
When R2 is high there's base current turning on the 2N3904. NO??? When the 3904 switches on there's a brief moment of current charging the IGBT Gate. That turns the IGBT on. When R2 goes low, the 3906 switches on and drains what little charge there may be in the gate of the IGBT. I think I have it right. If I have it wrong - rather than asking "Shouldn't your PNP and NPN be swapped?" I need to hear why it's backwards - if it is. I'm not challenging you on this - I am a basic hobbiest at best. I've never been any good with analog; so biasing a transistor is a mystery to me.
 
Last edited:

ElectricSpidey

Joined Dec 2, 2017
2,758
I think what Chris is saying is...that both of your transistors are in emitter follower mode.

I wouldn't worry about storage times in that mode because they will never saturate.
 

Beau Schwabe

Joined Nov 7, 2019
155
This is a common configuration as long as the input goes rail to rail of the supply to the NPN and PNP .... And you have some sort of return path on the two emitters which you don't.... The gate of the IGBT is capacitive and does not provide any kind of return path. This configuration is not suited for driving an IGBT or MOSFET
 

Chris65536

Joined Nov 11, 2019
270
When the 3904 switches on there's a brief moment of current charging the IGBT Gate. That turns the IGBT on.
I always use the collector of a BJT when switching something. Your design is an emitter follower, which is more useful for amplifying an analog signal. I am certainly not an expert, but this is my hunch based on experience.
 

Thread Starter

Tonyr1084

Joined Sep 24, 2015
7,853
OK then. I'll have to give it another go with it the other way around.

Thanks. It'll be a while before I get back to messing with the circuit - I am still in semi-cleanup mode. Plus I gotta go pick up a Bison Steak at the store.
 

crutschow

Joined Mar 14, 2008
34,285
No, it's correct the way you have it.
The transistors are acting as emitter followers which have a low output impedance so they can readily supply the MOSFET gate charge current, and are fast, since they don't suffer from the saturation storage delay that a common-emitter circuit does.

The emitters don't need a return path, as that's provided by the gate capacitance charge and discharge current.

If a transistor is getting hot, then it likely is incorrectly connected.
 

Thread Starter

Tonyr1084

Joined Sep 24, 2015
7,853
If a transistor is getting hot, then it likely is incorrectly connected.
Fairly confident it's wired the way it's drawn. Is it drawn wrong?

Been reading about the Baker Clamp and the RC speedup. Wondering if that might mitigate some of the overlap current that is possibly cooking the 3906.

HEY! Gotta run to the store. BBL
 

Chris65536

Joined Nov 11, 2019
270
The emitters don't need a return path, as that's provided by the gate capacitance charge and discharge current.
This is the part that I wasn't sure about. I imagined the PNP shutting off as the emitter voltage rises, causing it to rise more slowly and possibly causing problems in the IGBT. Honestly the whole emitter follower thing is kind of confusing to me, so I don't use it much.
 

ElectricSpidey

Joined Dec 2, 2017
2,758
My suggestion would be to reduce or eliminate R2 because emitter followers have a very high input impedance, and a bipolar 555 can probably handle the current.
 

Thread Starter

Tonyr1084

Joined Sep 24, 2015
7,853
Went with 1K at 12V because 12/1000 = .012A. IF I reduce it to 100Ω then that'd be 120 mA and the 555 is rated for 200 mA if I read the data sheet right. I'm not the expert on this but the reasoning for 1KΩ is the 1/10 rule I've sort of heard of. Eliminating R2 means there's a chance of drawing more than 200 mA if the 3906 comes on before the 3904 shuts down. It's only a guess, but maybe that's why the 3906 is getting substantially warmer than the 3904. Hardly measuring anything above ambient on the 3904.

So far I've been advised to change the circuit and to NOT change it. Right now I'm game to throwing a .01 µF cap across R2; based on some of the discussion I read in this thread started by @cmartinez 4 1/2 years ago.
 

crutschow

Joined Mar 14, 2008
34,285
Fairly confident it's wired the way it's drawn. Is it drawn wrong?
For the Post #1 circuit, no.
What is the switching frequency when it gets hot.
It certainly shouldn't get hot at low frequencies.
Have you double checked the transistor pinouts?
if the 3906 comes on before the 3904 shuts down.
Since there's over a volt difference between one transistor turning on and the other turning off, there's little chance of that happening.

Below is the LTspice simulation of an example circuit.
Note that the BJT drive voltage (red trace) is above the output voltage to the MOSFET gate
(yellow trace) when the voltage is rising and the NPN is on, and below the output when the voltage is falling and the PNP is on.

The green trace shows the charging current going into and out of the MOSFET gate.
Although the current is above the transistors steady-state rating, the current pulse is so short, that it should cause no damage to the devices.

1579653493686.png
 
Last edited:

Chris65536

Joined Nov 11, 2019
270
Since there's over a volt difference between one transistor turning on and the other turning off, there's little chance of that happening.
I now realize that my idea of swapping the BJTs would give a wide overlap where they are both on, so that was a bad idea. Thank you for this enlightening simulation! The gate voltage looks to rise and fall almost linearly, while I thought it would be a more gradual curve. I did expect the gate to be about 0.7V for off and V1-0.7 for on, and the simulation confirms that. Will the IGBT be fully off at 0.7V?
 

Thread Starter

Tonyr1084

Joined Sep 24, 2015
7,853
Have been having an issue with the overall circuit. Using a 100KΩ pot bought from Amazon - the darn things stink to high heaven. Temporarily I've swapped it with a floater pot that has been knocking around on my bench top for a few years. It's a 10KΩ pot - which means the frequency is a lot higher than originally planned. High frequency is most definitely not wanted or needed. I'm waiting on new pots. Meanwhile I'm looking through old boards with pots on them - from stereo systems with adjustable mic inputs - they have pots on them. I just need to find them. I have a little of everything - everywhere. Finding stuff can be a challenge at times. But I'm working on that issue too.

As for the pin-outs - yes, they've been checked and verified. They are as the data sheets suggest. These transistors came from Radio Shack. I dare not say "Not from China" - who knows from where Radio Shack gets their stuff.

The wave forms on pin 3 of the 555 is pretty square. No real anomalies to report on that. Coming from the transistors it's a little more rounded on the leading positive going edge. Still, nothing I'd think is that bad. The IGBT has a pretty sharp square waveform, so it looks like things are functioning as expected - least for say the 10K pot. That WILL be changed as soon as I can. Meanwhile I've been doing some housekeeping. GEEZ! I got a lot a small components laying loose around the bench. Is that normal? Or sick?
 

SamR

Joined Mar 19, 2019
5,031
I got a lot a small components laying loose around the bench. Is that normal? Or sick?
I keep a couple of small magnetic dishes on the bench to throw loose stuff into. And when they get full I FORCE myself to empty them, sort it all out, and put it back into storage where it came from. It's just gotta be done...
 
Top