Just go with increasing the noise gain as OBW5409 suggested in post #14.

It is nearly fool-proof and has a very high success rate.

If you find that the lower open loop gain is affecting DC accuracy you can install a capacitor in series with the shunting resistor (500 ohms in the given example). Use C = 1/(2 Pi R F) to find the capacitor necessary to have the gain reduction kick in at your desired frequency.

I tried a version of this without any improvement. Maybe I got it wrong. I will repeat the experiment.

 

That is a really bad idea. Solderless breadboards are about the worst thing you could use for any fast circuit. Ideal would be dead-bug style wiring on a solid copper plane, or at least a prototype board with square pads.

I don't expect to see any problems at only 10 MHz. I already have the triangle wave oscillator (without a fancy current source) running at 15 MHz. I was hoping for more like 50 MHz. With the correct high speed parts I think that is possible. For instance, right now I am using a slow ECL flip-flop and voltage comparators using a transistor array.

Yes, I said "slow ECL". The part has a typical prop delay of 3 ns.

 

I tried a version of this without any improvement. Maybe I got it wrong. I will repeat the experiment.

If I understand correctly, you tried my first attempt, in post #3.

See post #14, instead; that's the one DickCappels was referring to, and it solved the problem.

 

Continue to reduce the resistance. If you do not encounter stability we must look for other sources of the problem (the system is not oscillating but is receiving interference from an external source).

 

Continue to reduce the resistance. If you do not encounter stability we must look for other sources of the problem (the system is not oscillating but is receiving interference from an external source).

Right now this is only in simulation. I have been lazy in not plugging some parts onto a breadboard.

 

No, I have done the .AC analysis.

I keep meaning to test the circuit on a solderless breadboard but other things always seem to come up. I was in the process of laying out a proto PCB when I noticed a wrong resistor value. That led to my noticing the instability situation with increasing gain.

I always assume that Spice is much more likely to miss a problem than to report one that does not exist. I have seen some really weird things in Spice, however.

Hi,

Sorry to say but a solderless breadboard is out of the question for this kind of bandwidth, but if you just want to see if it works in general at slower speeds it may work to some degree. Those boards have other problems though that are not easy to define, so using one for a nice circuit like this is just asking for additional problems you dont need and may be hard to isolate.
Do yourself another favor and look for some links on good high frequency circuit layout techniques. It will save you a lot of time and energy

 

Sorry to say but a solderless breadboard is out of the question for this kind of bandwidth, but if you just want to see if it works in general at slower speeds it may work to some degree. Those boards have other problems though that are not easy to define, so using one for a nice circuit like this is just asking for additional problems you dont need and may be hard to isolate.
Do yourself another favor and look for some links on good high frequency circuit layout techniques. It will save you a lot of time and energy

Thanks for your concern. I have used SBB's since they were invented. I am very familiar with the limitations.

I am also familiar with good high frequency practices. I have done avalanche and "SCR" pulser circuits that have sub-nanosecond risetimes.
https://forum.allaboutcircuits.com/threads/fast-risetime-falltime-pulsers.124518/#post-816114

 

Thanks for your concern. I have used SBB's since they were invented. I am very familiar with the limitations.

I am also familiar with good high frequency practices. I have done avalanche and "SCR" pulser circuits that have sub-nanosecond risetimes.
https://forum.allaboutcircuits.com/threads/fast-risetime-falltime-pulsers.124518/#post-816114

Hi,

Oh well what did you intend to use the solderless breadboard for then?

The first logic used vacuum tubes in a manner such as RTL, but i read somewhere that there were also logic circuits that were implemented as just DRL (diode resistor logic) where the diodes where vacuum tubes. I've used diode resistor logic in the past myself but using silicon diodes, starting probably early 1970's.

 

Oh well what did you intend to use the solderless breadboard for then?

The first logic used vacuum tubes in a manner such as RTL, but i read somewhere that there were also logic circuits that were implemented as just DRL (diode resistor logic) where the diodes where vacuum tubes. I've used diode resistor logic in the past myself but using silicon diodes, starting probably early 1970's.

I use the SBB to check the logic of the design. In other words; to check for obvious design errors such as the wrong feedback phase.

I have been reading about the ABC (Atanasoff Berry Computer). It used "RRT" (Resistor Resistor Tube) logic. The logic inputs were summed into the grid of a triode. See page 15 of the attached PDF.

The illustrations are from the book The First Electronic Computer: The Atanasoff Story by Alice R. Burks and Arthur W. Burks
https://www.amazon.com/First-Electronic-Computer-Atanasoff-Story/dp/0472081047

 

Success! The current source is now stable -- at least in simulation .
I can now put the current source on the SBB prototype. I will get back to you on the results.
I will add the compensation components to the PCB I am working on, as well.

Here is a picture of the function generator on an SBB. This is the circuit that oscillates at 15 MHz. It is similar in operation to the old Intersil ICL8038. Note that the circuit is compact and never uses a wire longer than is needed

The IC's are (from left to right) a LM6181 current feedback op-amp, CA3102 transistor array and an MC10135 ECL flip-flop.
The IC seen on end is used to generate -5 volts from the 5 volt input power.
The parts on the SBB are what I had on hand...
A couple of 2N3640's are used to switch the polarity of the current used to charge and discharge the timing cap. They do not switch real fast and have a lot of capacitance.
The LM6181 buffers the timing cap voltage. It is a bit slow and has terrible input bias current.
The CA3102 is used to make a crude dual voltage comparator. It has low gain and is somewhat slow.
The MC10135 is an old ECL flip-flop that has a huge propagation delay (for ECL).

Since this prototype was built, I have made improvements to the design that should speed it up a bit and use fewer parts -- a win-win.
I am going to add the improvements to the prototype when I add the current source circuit you-all helped me get working.

 

Success! The current source is now stable -- at least in simulation .

Just curious, how did you end up stabilizing it?

 

Just curious, how did you end up stabilizing it?

Duh! How could I have forgotten to tell? Thanks, again, for your help.

A 500 ohm resistor in series with a 1000 pf cap between the inputs of the op-amp.
Still some nasty glitches but dealing with these will have to wait for real parts on a PCB.

 

Duh! How could I have forgotten to tell? Thanks, again, for your help. A 500 ohm resistor in series with a 1000 pf cap between the inputs of the op-amp.

Excellent!

 

Here is my version of the scheme

 

Here is my version of the scheme
View attachment 137556

Hi,

That's exactly the way i was describing and that's how it's usually done.
If i get a chance i'll look into the math behind this but i have something a little pressing today that needs attention first.

Another somewhat intuitive view is that the transistor gain is uncompensated even though the op amp gain might be. That makes the whole thing uncompensated.
To see what a pain this can be we only have to look at some low drop out positive voltage regulators that use a PNP output stage. One of the problems is keeping it compensated for various types of loads.

 

I did not claim to be original. In the scheme there is an additional condenser, which carries out additional filtration (removes spike at the fronts). I just painted a practical implementation. First I took the original circuit and using RC-schemes, I achieved the lack of generation. But the circuit worked poorly at a frequency of 100 kHz. Instead of rectangular pulses, triangles were obtained.

 

Here is my version of the scheme

Thanks. That may very well work in my design. I will look at it later.

Drat. I just realized I will have to add a third PNP in my design. There is also a negative current source like the positive one we have been fixing. Adding the extra PNP may not sound like a problem but it really is. Ultra high frequency PNP's are rare.

 

Here is my version of the scheme

Drat. I just realized I will have to add a third PNP in my design. There is also a negative current source like the positive one we have been fixing. Adding the extra PNP may not sound like a problem but it really is. Ultra high frequency PNP's are rare.

Oops. Now I realize that the PNP will be in the low frequency path. Not a big deal other than PCB space.

 

I did not claim to be original. In the scheme there is an additional condenser, which carries out additional filtration (removes spike at the fronts). I just painted a practical implementation. First I took the original circuit and using RC-schemes, I achieved the lack of generation. But the circuit worked poorly at a frequency of 100 kHz. Instead of rectangular pulses, triangles were obtained.

Hi,

Oh my comment was not a comment on your originality or lack thereof, it was actually a complement that you did it the way i think is the best way and most common.

 

Oh my comment was not a comment on your originality or lack thereof, it was actually a complement that you did it the way i think is the best way and most common.

Thanks MrAl. I apologize for not understanding what you were saying until @Bordodynov posted his example circuit.

I have decided on adding the transistor rather than compensating the op-amp. This lets me change the type of op-amp I use without having to redo the compensation values. For instance, I can test with a slow op-amp that I have on hand and then upgrade to a faster op-amp when I get it.

I can now update the PCB layout with the changes and do some etching.

I learned two things. How to compensate an op-amp with added gain and another way to do the current source.