Saturation of the preamp

[Tinkerer]

I found a major hurdle in reducing the first sample delay, That is the saturation of the op amp.
Question: would it help to run the op amp on +- 15 V instead of +- 5 V?

How do I read the spec sheet for info on transient response and saturation ?
Tinkerer

[Ferric Toes]

Tinkerer

Provided you are using a reasonably fast opamp, most of the saturation time is caused by the decaying back emf from the coil. This is first clipped by the avalanche breakdown of the Mosfet and then decays exponentially from, maybe a few hundred volts down to the 600mV clipping level of the protection diodes. At this point the amplifier is still saturated hard, and if its gain is, say, 500x, you have still got to wait till it has dropped to less than 10mV (+ & -5V supplies) so that the amplifier can then come out of saturation. All this can take several microseconds, with the amplifier recovery adding less than a uS. All this still depends on the damping being optimised. Anything less, or more, and the time can be stretched out even further.

Opamp data is only a rough guide for this application. As you can see from my recent posts, two that looked good on paper, failed to perform.

Eric.

[Tinkerer]

Avalanche break down of the Mosfet

Thanks for the explanation, Ferric Toes,
So if I understand right, the avalanche diode voltage could be designed into the damping circuit. I mean by choosing a specific voltage lets say 300 V, the flyback peak would be clipped at that voltage, from there on the damping resistor does its job.
If I choose a 800 V voltage and the flyback peak only reaches, say 600 V, the avalanche diode does not enter into play. Is that right?
Tinkerer

[Ferric Toes]

Thanks for the explanation, Ferric Toes,
So if I understand right, the avalanche diode voltage could be designed into the damping circuit. I mean by choosing a specific voltage lets say 300 V, the flyback peak would be clipped at that voltage, from there on the damping resistor does its job.
If I choose a 800 V voltage and the flyback peak only reaches, say 600 V, the avalanche diode does not enter into play. Is that right?
Tinkerer

Correct on both points. The avalanche clipping is a characteristic of the Mosfet. e.g. an IRF740 avalanches at 400V, and almost looks like a short circuit across the coil. When the voltage has dropped below 400, then the damping resistor takes over. A higher voltage Mosfet will let the voltage rise higher, but as you say, it may not reach the clipping point. It is then all down to the damping resistor to dissipate the energy. You can get a faster switch off with a high voltage device, but the peak current in the opamp protection diode also gets a bit high. If you use a 1k opamp input resistor and the voltage peaks at 600V, then you have a current peak of 600mA in the 1N4148, or whatever you are using. The diode data gives a maximum non-repetititive current of 4A for 1uS, so it is not clear how far you can push it with a repetitive pulse. I can't remember when I last blew one though.

Eric.

[nncunew]

And ?

Sorry, but i have a question refered to this suject.

And how you select the optimun damper resistor ???

There are two resistor, one conect in paralell with the coil and other in serie with the diodes... how i can obtein the optimun values for this two resistors ???

Sorry for my bad english

[email protected]

[Tinkerer]

different aproach to clipping

Correct on both points. The avalanche clipping is a characteristic of the Mosfet. e.g. an IRF740 avalanches at 400V, and almost looks like a short circuit across the coil. When the voltage has dropped below 400, then the damping resistor takes over. A higher voltage Mosfet will let the voltage rise higher, but as you say, it may not reach the clipping point. It is then all down to the damping resistor to dissipate the energy. You can get a faster switch off with a high voltage device, but the peak current in the opamp protection diode also gets a bit high. If you use a 1k opamp input resistor and the voltage peaks at 600V, then you have a current peak of 600mA in the 1N4148, or whatever you are using. The diode data gives a maximum non-repetititive current of 4A for 1uS, so it is not clear how far you can push it with a repetitive pulse. I can't remember when I last blew one though.

Eric.

Good you mention the problems with the clipping diodes.
Is there anything to be gained by using Schottky diodes? Or using transistor junctions as diodes, or biasing the diodes or transistors to clipp at a lower level , say 10 mV?
Tinkerer

[Sean Goddard]

Hi All,

Regarding the use of diodes to clip the waveform, I saw an interesting idea involving back to back transistors. This arrangement seemed to react MUCH faster than the diodes as it was "self feeding" in some way.

Can't for the life of me remember how the circuit looked, but it used the B-E junction ONLY of one of the transistors and DIDN'T saturate!!

If I can find it, I'll post it here.

Eric, any ideas?

[Ferric Toes]

Hi Sean,

I have an article too, somewhere. As well as just using one junction, there was also a "transdiode" connection, where the base was strapped to the collector. Can't remember what the relative merits were.

The early PI's I worked on used a couple of transistor junctions, but later on I found the 1N4148/1N914 to do the job adequately. It would be interesting to look again at a transistor junction, particularly for clipping the back emf spike, which is the important one.

Eric.

[Ferric Toes]

Bounding, clamping techniques improve circuit performance.

Found the article. It is in EDN Nov.10th 1983, by the famous Robert Pease, title as above. Tells you nearly all you need to know.

Eric.

[Reg]

Hi Eric,

I am not able to find the R Pease article about the transistors as bounding diodes on the net. When you get a chance, can you summarize it or scan it and attach as a pic file.

Thanks,

Reg

[Reg]

Hi Sean,

George Payne used transistors as diodes in some of his designs. He used two transistors back to back, each with the base tied to the collector.

One patent by Hirschi and Payne shows them. I think it is 4470015 is one that displays them in use. Just look on sheet 5 of 10. His US patent number 4507612 also shows them on pages 6 and 7.

Reg

[Ferric Toes]

Hi Reg and All,

Here are a couple of extracts from the article I mentioned. There are too many pages to scan and post, but these two parts give some relevant information for choosing clipping diodes. The article is 1983, so there may be other sources of later info.

Eric.

[JC1]

EE

Hi Guys,

And if you look at the schematic diagram of the 5534 opamp you will see the clamp transistors on the input. Also see other diode transistors through out the circuit.

Alot of IC diodes are formed by making transistors in the silicon (since they are doing that anyway) and then short the base to the collector to make a diode.

[Ferric Toes]

Hi JC1,

I've actually run a 5534 without any external clamp diodes. It works fine on my low power detectors, where I am pulsing 0.1A through the coil, but on a standard unit with say an IRF740 pulsing at a few amps, the internal diodes blew. Or at least one of them did. This goes back to my point about peak diode current on the back emf spike.

Eric.