On Tue, Apr 1, 2025 at 05:51 AM, Christopher Paul wrote:

It’s unclear to me which of the four separate circuits you are referring to, or why you are suggesting a reconfiguration and what goal you are pursuing.

?

My only goal is to try to help. But the circuit doesn't make sense. I think you are comparing the behavior of the groups of circuits. But what is the final circuit supposed to do?

?

Is the 1 ohm resistor the load?

On Tue, Apr 1, 2025 at 08:51 AM, Christopher Paul wrote:

It’s unclear to me which of the four separate circuits you are referring to, or why you are suggesting a reconfiguration and what goal you are pursuing. All are configured as I intended. They are independent of one another except for sharing a power supply.

I think eT was getting confused about the fact that there were two current sources, which he thought both fed current into the string of diodes.? They don't, of course.? (Even if they did, they do not conflict or fight.? Current sources can be directly added without problem.)

?

The op-amp+MOSFET current source controls the 100-times-larger LM317A current source, by controlling its voltage sensing pins (the voltage between its OUT and ADJ pins).? We don't know exactly why Chris has it arranged that way, but I'm guessing it is either to toggle the LM317's current source on and off, or to adjust its current level.

Although we still don’t know, the simulation completes and runs (acceptably) slowly with Andy’s suggestion of the alternate solver.

You can speed up the simulations.? With the Alternate solver, or the pull-down resistor, LTspice wastes a lot of time trying (and failing) to find the initial operating point.? By examining the results in the Error Log file, you can see which of its algorithms failed, and then add appropriate .OPTIONS commands to skip those algorithms and save time.? For example:

? ? .OPTIONS NoOpIter GminSteps=0 SrcSteps=0

makes it go straight to the Pseudo-Tran algorithm, which is the only one that succeeded with just the Alternate solver.? If I remember correctly, the Alternate Source Stepping algorithm succeeded after adding the pulldown resistor, so omit "SrcSteps=0" in that case.

?

See LTspice's Help > LTspice > Dot Commands > .OP -- Find the DC Operating Point.? The information on that page applies to all simulations, not only the ones that have an ".OP" simulation command.

?

I don't know what causes the first two or three algorithms to fail.? Sometimes that just happens and we need to live with it.? That is why LTspice has four algorithms, in case some of them fail.

?

Andy

?

On Tue, Apr 1, 2025 at 05:03 AM, Tony Casey wrote:

Threaded discussion list view can be enabled in other email clients besides Thunderbird, including Gmail, Kmail, Outlook and Windows Mail. In Gmail, Outlook and Windows Mail it is called "Conversation" view, but it needs configuring in most clients and is best used when filters or rules are applied to your inbox(es) to move certain messages to other folders.

Gmail's "Conversation" view is very much NOT the same as Thunderbird's Threaded view.? Gmail ignores the same header line that identifies message threading that Thunderbird uses.? When someone hijacks a topic and changes the Subject line, Gmail always puts it into a separate topic in Gmail, unlike what Thunderbird does.? (Actually I think that is a combination of both the Subject line and another hidden header line.? But it definitely differs from Thunderbird's view.)

?

My recollection is that Outlook also ignored threading (conversations were grouped but not threaded), and used something like the Subject line to identify conversations.? But I have not used Outlook in many years.

?

My first home internet email reader had a proper threaded view, which I really missed when I eventually changed to different email programs.? And still miss.

?

I am sometimes guilty of replying to the wrong topic in a thread.? Sometimes I pick a random topic to reply to (often it's either the first or the last), which probably does not look right in a properly threaded message view.? I should stop doing that.

?

Andy

?

On Mon, Mar 31, 2025 at 07:59 PM, eetech00 wrote:

I haven't tried this but, then the anode of the top diode should connect to the LM317 current regulator output and the mosfet drain should connect to the cathode of the bottom diode in the string(?)

I think that would not make sense, because there's about a 100 to 1 difference in the two current limiters.

?

The op-amp+MOSFET part is a (logical) input into the '317 regulator.? NOT meant to drive the diode load, and it does not drive the diodes at all.

?

Andy

?

I haven't tried this but, then the anode of the top diode should connect to the LM317 current regulator output and the mosfet drain should connect to the cathode of the bottom diode in the string(?)

The op-amp+FET sinks?current downwards.

?

The LM317A sources current to the diodes, plus the FET.? They don't compete.

?

There are 2 orders of magnitude difference between the two current regulators.

?

I think the FET sink sets the drop across a resistor, 'tuning' the much greater current regulated by the '317.

?

Andy

?

On Mon, Mar 31, 2025 at 02:14 PM, Christopher Paul wrote:

eetech00, I’m afraid I don’t understand.

?

From what I can see, each of the three diode strings is driven by a single current regulator. Maybe you could you describe your concerns using component reference designators?

?

?

?

You have three groups of circuits, but I'll refer to one group.

The opamp and mosfet forms a basic Volt to current converter and its controlling current thru the diodes.

The LM317 is also configured as a current regulator and it also (trying) control current thru the diodes.

That's what I think is causing the simulation convergence problems.

Chris

?

The circuit run fine if its connected correctly.

The way it is connected, the two current regulators are in conflict with each other, since they are both trying to regulate current to the same load (the diodes). Use two separated current loads, one for each regulator.

Hello everyone,

First of all, I appreciate your comments, suggestions, and circuits. I am reading the papers and will simulate what you suggested. I work in a laboratory where we are characterizing some biological samples, and they exhibit Memristor-like behavior or something similar.

A new file has been uploaded: ? ? ? ? ? ? ? ? ? ? ?Negative_Resistance_Ckt.asc.zip ? ? ? ? ??? This is a generic (architectural) schematic with no component values but it IS an LTspice file. Pick a zener (typically around 5V), an NPN transistor (I use 2N3904), a PNP transistor (I use 2N3906), ?I use a few KOhm for R1, R2, R3, R4. ?I use a few hundred ohms for R5. R4 should not be so large that Q1 saturates at peak current. Avoid reverse base-emitter breakdown of Q1 at maximum input voltage.

?

Jim Wagner

Oregon Research Electronics

So, sorry - that drawing has been bad for lots of years! ?Q2 should be PNP! ?I will correct it. Don't want bad stuff out there. Should I remove the incorrect post?

A correctly made model would not care whether the voltages are +/-15, or +30, or (+20 & -10) volts.? It's all just relative.? From the op-amp's perspective, the only difference is where the external ground voltage happens to be.

?

Of course there are badly constructed SPICE models too, that do the wrong thing.? Hopefully those are few now.

?

Andy

?

On Sat, Mar 29, 2025 at 09:38 PM, Jim Wagner wrote:

I think I have the circuit you need. It is circa-1965 and uses 3-4 bipolar transistors, a zener diode, and a handful of resistors. ...

Jim,

?

Your schematic in the photo you uploaded shows two NPN transistors, but the description that accompanies it mentions "2N3904/6" which implies that one of them is a PNP.? I have not tried (nor studied the circuit in detail), but I wondered if Q2 is supposed to operate in reverse breakdown mode so it is an NPN, or if it was supposed to be a PNP.

?

Did you draw that schematic, or was it originally drawn with two NPNs, and is one of them wrong?

?

Andy

?

Sorry, I also should have said this:? Ground the Emitter and connect the negative collector voltage source to the transistor's collector.

?

Obviously, the graph you uploaded needs to be inverted (or plotted with negative voltages going right and negative currents going up) when using a PNP.

?

Andy

?

On Mon, Mar 31, 2025 at 07:27 AM, john23 wrote:

Is a there a way by clever type of sweep to recreate the basic characteristics plot of the attached PNP plot?

Sweep the collector voltage.

?

Sweep the base current over a number of steps.

?

?

Andy

?