Improving Transistor Switching Speed

In this video I demonstrate how a simple transistor switching circuit can be improved so that it will operate at higher frequency.

[@HyperspacePirate]

Thanks, this has been really helpful. One question i had, though, was how does this relate to the "Delay Time / Rise Time / Storage Time / Fall Time" figures provided in the BJT's datasheet? For example, I'm driving a 2N3904 at 10 MHz , and the delay time + fall time + rise time per the datasheet add up to 120 nS, but the half-cycle time of 10 MHz is 50 nS and I'm still getting switching, which doesn't seem like it should be possible. Am I misunderstanding those time figures?

[@abhaynathani9541]

Thanks you a lot! ❤

[@Victurf]

Super! Merci beaucoup!

[@FambiTech]

may i know which transistor you are using??

[@magiclay]

what transistor are you using?

[@robinharris4706]

Great video - thanks Jerry. Been experimenting with non switching time measurement and about to try speeding those up with some ideas like yours. I'd like to replicate your demo - could you tell me the resistor values on the base please, especially the one to ground? I haven't been using that so far.

[@leyasep5919]

One possibility for a kit is to provide the pads for the extra parts and leave the user the choice to populate the diodes (and capacitors, in parallel with the series base resistor).

[@JerryWalker001]

Unfortunately this would double the size of the boards and they are already fairly large and this would add £100.00 to the kit price just for the board cost. The kit it not intended to run at high speed as this does not allow the operation to be seen. It was never intended to run beyond 2KHz (It will actually run up to 250KHz) so the current design fully meets the intended purpose.

[@jontscott]

That is interesting, thank you for sharing. One question that crossed my mind is if you swap the bipolar for a mosfet (with a low gate capacitance not some big power fet) would that greatly improve the speed? Also if you chain say 3 of those 'inverters' back to back I would guess that would cut the top speed even more? Would the diodes have as much of an effect when stages are chained since subsequent stages would no longer have the low impedance drive?

[@JerryWalker001]

Unfortunately the circuits would not work with mosfets although there is no reason you could not design a system using mosfets but the general design would need to be very different as the bipolar transistors have benefits in these simple gate designs. In particular designing multi input gates would be much more difficult. The processor works up to around 250KHz which may sound odd considering that I show a single stage working up to 6MHz but as you correctly say there are more than one stage in the processor circuits. Some have 8 gate stages and so the total delay is the combined delay and so the limiting switch rate is correspondingly lower.

[@nathanthomas9395]

what is the name of the signal generator?

[@reimundolopez1198]

Y como pongo eso, en un transistor pnp?

[@FireballXL55]

Very nice demo. For high power devices like used in flyback circuits they reverse bias the base region so it zeners and discharges the base region either by a base drive transformer or a large capacitor. Obviously you cant do that in a multi transistor circuit.

[@JerryWalker001]

Yes I showed a similar technique using a capacitor in a previous video. That technique increases the 'drive' rather than clamping the switch point which cannot be done in a flyback as that would destroy the device, In the first modification in this video uses the same drive increase technique and in the second modification it uses a clamp (actually gain control) method to give better results. They are useful methods in some circuits but as you say it was not really an option for the processor kit due to the number of transistors used and was also not needed as the processor was only supposed to be used up to around 2KHz. It actually works up to around 250KHz.

[@robinharris4706]

I've used this circuit with two diodes limiting saturation and it makes a huge difference with non-switching devices like BC108; a significant improvement but not as much with 2N2222 and 2N3904. I tried to find more details on how it works but haven't found it anywhere else. Does this method have a name or can you point me to anywhere that explains the mode of action?

[@JerryWalker001]

The main limiting factor in bipolar switching speeds is the junction capacitance when the device is driven into saturation. This varies from device to device and high speed devices are generally made with smaller junctions which is why they are often more 'fragile'. What you need to do to speed up the slower devices is to prevent them being driven into saturation (which is the purpose of the diodes) and increase the junction turn off time by bleeding off the charge (which is what the resistor is for). The values selected depend on transistor type and what trade off's you are willing to make. You will get less performance improvement with switching types because they are already optimised and so the returns are smaller.

[@robinharris4706]

@@JerryWalker001 Thanks for the additional information. I'll experiment with a 'bleed' resistor on the BC108 family that are slow switchers then.

[@Paul_Bearden]

You could also use a mosfet like a 2N7000.

[@JerryWalker001]

Unfortunately the 2N7000 will not work in something like the transistor processor. It is also not a bipolar transistor which is the real point of this video. You could for example also replace it with something like a 7404 but the point is how to make a bipolar transistor switch faster and not how to replace it with something else.

[@Paul_Bearden]

@@JerryWalker001 Can you explain specifically why that is?

[@JerryWalker001]

@@Paul_Bearden Which part of my reply are you questioning? If it that it will not work then you may want to purchase the book and try to copy the circuits using a 2N7000. They will not work because bipolar transistors work very differently to a 2N7000 and you cannot connect something like a 2N7000 in circuits in the same way as bipolar transistors. If it is that it is not a bipolar transistor then it is not a bipolar transistor which is specifically what the video was about. As I said you could replace the bipolar transistor with something faster but that is not really making a bipolar transistor switch faster. There are other reasons the 2N7000 would also be unsuitable for something such as the transistor processor. For example they are at least 5 times the cost and as there are 6000 of them in the processor (with RAM) then the cost would be very high. You could of course redesign the circuits to use something like the 2N7000 but as the processor and the video specifically relates to bipolar transistors then a 2N7000 does not qualify.

[@Paul_Bearden]

@@JerryWalker001 You are incorrect. MOSFETS work exactly the same as BJT transistors, except that BJT transistors vary the output voltage according to the input voltage, while a MOSFET varies the current while passing any voltage. Also, I have just ordered 40x 2N7000 on ebay for $5 free shipping.

[@JerryWalker001]

@@Paul_Bearden That is absolutely NOT the case. I suggest you try it. Bipolar transistors and MOSFETS are VERY different in complex circuits. They may work in a similar manner in a simple switching circuit but not when you connect them directly together. Instead of simply contradicting me I suggest you try it. As I said they are MUCH more expensive than bipolar devices. Your pricing results in a total cost for just the 2N7000's of $750.00. You can buy 6000 bipolar devices for around $70.00 (11 times cheaper) so I stand by what I said. I sell the entire kit including boards and all parts for less than $750.00.

[@rlgrlg-oh6cc]

The diode from the collector looks backwards in your schematic. When the collector goes below a certain voltage, it should pull the base drive down. The way you have it, when the base voltage is low, it will hold the collector low.

[@JerryWalker001]

No it is correct. It is not a voltage clamp it is a bypass diode circuit which controls the gain of the transistor as it reaches saturation. I inserted the diodes exactly as shown in the schematics and then showed the circuit working.