apzx

Small Signal Diode Differences
« on: June 04, 2020, 09:28:53 AM »
Perhaps I am a bit dense and not understanding something correctly, but regardless I am kind of at a loss and Google didn't turn up too much. Probably because I am not sure on what it is I am looking for here.

So, I am playing around on the simulator (dangerous I know) and in particular looking at the NTP M100 DOA. Mainly because I am trying to better understand how it works, and it is not an overly complicated design. I've got a few of the parts down, but not everything. However, my question does not inherently pertain to the DOA itself. I remember Sam Groner mentioning in a thread some suggestion of transistors to use in the circuit. And the circuit looks like it'd run great with them. However, I decided to try substituting different transistors in the design and really just took note of the differences that happened. Then I got curious to see if the diodes made any appreciable difference, and much to my surprise it made a considerably larger change than I thought they would.

I initially started with the 1N4148, and then I went over to Mouser to see what other small signal diodes were readily available and I ended up trying 1N3595s, 1N457s, and BAV20s. Why these in particular? No real reason, I just wanted a small handful to mess around with. Though I am not sure what it is going on here honestly. Like I can see with 1N4148s that idle the DOA draws about 5.8mA. However, change that to the 1N3595 and it that moves to about 10.3mA. With the BAV20 it is about 8.7mA. With 1N457s it is more or less the same as the 1N4148.

So, what is going on here? My guess is that it has do with the forward voltage of the diodes wrt to the current, and since the DOA is biased with them then by happenstance of the transistors and changing their operating currents makes them behave a bit differently. The main way that I was comparing differences is primarily in the stability of the DOA. And even though the 1N3595 is a slower diode than the 1N4148 the 1N3595 had more gain and phase margin at higher frequencies (~2MHz increase).

I get that there are different diodes for a reason and I am really just trying to understand a bit better the actual data I am seeing here. I doubt in practice this really makes much of a difference, but I found it curious enough that I am surprised I did not find much regarding the actual differences in signal diodes. Again probably because I do not know what to search for.


JohnRoberts

Re: Small Signal Diode Differences
« Reply #1 on: June 04, 2020, 10:03:36 AM »
If it didn't make any difference there wouldn't be so many... I generally use small switching dioded line 1n914, and 1n4148 interchangeable for non critical applications.

For TMI back at Peavey we had a house part number dual diode sourced from motorola for use in biasing class A current for power amp output stages.

JR
« Last Edit: June 04, 2020, 06:11:12 PM by JohnRoberts »
Don't only half-ass tune your drums. Visit https://circularscience.com to hear what properly "cleared" drums sound like.

Newmarket

Re: Small Signal Diode Differences
« Reply #2 on: June 04, 2020, 06:03:57 PM »
Just a note to suggest you might want to look at connecting jfets as low leakage diodes.

Re: Small Signal Diode Differences
« Reply #3 on: June 04, 2020, 10:37:22 PM »
Just a note to suggest you might want to look at connecting jfets as low leakage diodes.

+1

Also, when it comes to diodes, they have different characteristics, forward voltage is not the same,  rise and decay times are different, also the capacitance of the diode is different, this is specially important when used reverse biased, check out Merlin's article on electronic design https://www.electronicdesign.com/technologies/analog/article/21131760/input-protection-for-lowdistortion-opamp-circuits, he compares the traditional 1N4148 with the BAV99 and the distortion differences are staggering, so yeah, not all diodes are created equal.

If you are using diodes to bias the output stage of the opamp, you may also want to consider a VBE multiplier, this way you can trim exactly the ammount of current needed in the output stage and depend less on diode forward voltage.

PRR

Re: Small Signal Diode Differences
« Reply #4 on: June 04, 2020, 10:44:40 PM »
> even though the 1N3595 is a slower diode than the 1N4148 the 1N3595 had more gain and phase margin at higher frequencies

It's not the diodes. It is the amplifier idle current set by diode voltage.

You could find a 0.678V battery and get the same performance (for a day or two). As said, a Vbe multiplier puts YOU in control. (However it is astonishing how often '4148 diodes set medium-size BJTs to a happy zone.)

apzx

Re: Small Signal Diode Differences
« Reply #5 on: June 05, 2020, 05:26:50 AM »
> It's not the diodes. It is the amplifier idle current set by diode voltage.

You could find a 0.678V battery and get the same performance (for a day or two). As said, a Vbe multiplier puts YOU in control. (However it is astonishing how often '4148 diodes set medium-size BJTs to a happy zone.)


So, if I am understanding correctly here running the transistors with greater Iq allows them to operate (I do not know the word here sorry :( ) more efficiently or operate at a higher frequency? Playing around with the circuit in the simulator the LTP is biased from a PNP current source. Changing the 1N4148s to 1N3595s results in an increase of the bias current of about 54.5uA for the LTP. Now, it looks like by increasing the nominal Iq of the LTP a little more response is achieved.

I also swapped out the two output biasing diodes for a Vbe multiplier in the sim and it again looks like if I increase the Iq there as well I notice a similar thing happening. That is a little more response is achieved. I do kind of wish I knew the word for this. Regardless the Vbe multiplier is really cool because it is incredibly easy to simply vary the Iq of the output stage by changing one resistor value. I am going to assume, without having looked for it in particular, that there is a limit to how far things can be pushed this way before other things start getting in the way, correct?

JohnRoberts

Re: Small Signal Diode Differences
« Reply #6 on: June 05, 2020, 10:03:32 AM »
> It's not the diodes. It is the amplifier idle current set by diode voltage.

You could find a 0.678V battery and get the same performance (for a day or two). As said, a Vbe multiplier puts YOU in control. (However it is astonishing how often '4148 diodes set medium-size BJTs to a happy zone.)


So, if I am understanding correctly here running the transistors with greater Iq allows them to operate (I do not know the word here sorry :( ) more efficiently or operate at a higher frequency? Playing around with the circuit in the simulator the LTP is biased from a PNP current source. Changing the 1N4148s to 1N3595s results in an increase of the bias current of about 54.5uA for the LTP. Now, it looks like by increasing the nominal Iq of the LTP a little more response is achieved.
I am not sure we are all on the same page here but the class A reference here is to the class A fraction of class AB power amp output stages. In common collector output stages using a NPN to source current and PNP to sink current, there are turn on and off delays when the output current transitions from sourcing to sinking. This delay causes a well know distorion mechanism called "crossover distortion". It is most noticeable at low levels and high frequency because of the time delay component.  A relatively simple remedy for crossover distortion is to leave both devices slightly turned on, this results in a much cleaner transition between sinking and sourcing current, but has a cost. Since both power devices are turned on with the full power supply across them, dissipation can easily become an issue. Too much class A current can release the magic smoke inside power transistors.

There are many different ways to manage this some as simple as using special well behaved diodes in the base drive circuit, to Vbe multipliers with trim pots.

 
Quote
I also swapped out the two output biasing diodes for a Vbe multiplier in the sim and it again looks like if I increase the Iq there as well I notice a similar thing happening. That is a little more response is achieved. I do kind of wish I knew the word for this. Regardless the Vbe multiplier is really cool because it is incredibly easy to simply vary the Iq of the output stage by changing one resistor value. I am going to assume, without having looked for it in particular, that there is a limit to how far things can be pushed this way before other things start getting in the way, correct?

A good starting point is around 25mA class A current. Generally you can measure the voltage drop across power device emitter resistors. If you don't have a distortion analyzer handy, you can look at the output on a scope with a small voltage, HF sine wave. You won't easily hear the crossover distortion on a HF sine wave, since the distorion products are at even higher multiples.

JR
Don't only half-ass tune your drums. Visit https://circularscience.com to hear what properly "cleared" drums sound like.

PRR

Re: Small Signal Diode Differences
« Reply #7 on: June 05, 2020, 02:43:43 PM »
> So, if I am understanding correctly here running the transistors with greater Iq allows them to operate (I do not know the word here sorry :( ) more efficiently or operate at a higher frequency?

Re: Small Signal Diode Differences
« Reply #8 on: June 05, 2020, 05:13:58 PM »
Having a higher current in the output stage lowers distortion because the Class A region of operation is extended, if the current is high enough, the devices might never turn off and remain forever in Class A, however, like John said, this comes at a price, and that is higher heat dissipation of the output transistors. If you want to bias into Class AB, there is an optimum output current  which some authors call the "Oliver" condition https://www.hpl.hp.com/hpjournal/pdfs/IssuePDFs/1971-02.pdf page 11 describes this.

the "Oliver" condition
« Reply #9 on: June 05, 2020, 05:40:24 PM »
and what class of amplifier is Dr. Oliver referring to ??


apzx

Re: Small Signal Diode Differences
« Reply #10 on: June 05, 2020, 07:29:29 PM »
Well this turned into a deeper rabbit hole than I thought and it went in a direction that I was not expecting, which is a good thing!

JR,
Yeah we're not. So, I'm going to try and explain what it is that initially sparked my question and then what it is that I have learned.

I was playing around with the NTP M100 DOA in the simulator. Mainly, because I want to know more about how opamps work internally, and this particular one is not overly complicated in design. I have figured out bits and bobs of it, but not all of it, and that is okay because sometimes I'm dense and need some time to process what is going on :)

Now, in order to make comparisons of changes in the circuit (transistors, resistors, diodes, etc...) I needed to opt for a simple test, and the test I opted for is open loop stability performance. Initially, per an old post by Samuel Groner I started with his proposed transistors (mainly 2N4403s, a MPSA18 for the VAS, and MJE180/170 or BD139/140 for the output). Then PRR suggested a bit later in the same thread to use a high gain PNP for the LTP because they are run with a fair bit of current. So, I swapped the input 2N4403s for 2N5087s, and I noticed that there was an increase in the open loop stability (i.e. extended response). Then I started going further and trying different small signal transistors throughout the circuit, and I was able to extend the response of DOA just by doing that. Then one day I figured why not try swapping out the diodes, because I incorrectly assumed that they did not make a big difference. This mainly stemmed from looking at a bunch of schematics and seeing 1N4148s or 1N914s and equivalents being used. In my own experience when I need a small signal diode that is what I use as well. So, when I ran the simulation with the 1N3595 (just as an example) and I noticed a rather large increase in the response of the DOA I was rather surprised. Hence spurring my question of what are the differences in Small Signal Diodes because by looking at the datasheet of the 1N3595 vs the 1N4148. The 1N3595 has more capacitance and is a slower diode.

So, from PRR's post instead of looking at the diodes as diodes I looked at them as current sources. The M100 uses a current source that I have played with before. Now, the overall current flowing through the diode string really does not change much, perhaps a 100uA or so. However, the current flowing through the LTP varies quite considerably, and at this point I was not sure what graph or what spec in a datasheet to look at to determine what is actually happening to understand what is actually going on. I also know that the output stage of the M100 is Class AB, sure I do not know how far exactly it stays in Class A, but that is a hurdle I will come to at a later point ;) Right now I am just trying to see how far I can extend the response while not going overboard on the idle current of the DOA.

PRR,

That is just the ticket! You're a gentleman and a scholar! I sort of feel like I have reached the top of the hill and there is now a few different roads to go down and explore.

And seriously thank you all for bearing with me, seriously I can be dense sometimes.

PRR

Re: Small Signal Diode Differences
« Reply #11 on: June 05, 2020, 08:06:30 PM »
> If you want to bias into Class AB, there is an optimum output current  which some authors call the "Oliver" condition https://www.hpl.hp.com/hpjournal/pdfs/IssuePDFs/1971-02.pdf page 11 describes this.

and what class of amplifier is Dr. Oliver referring to ??

Seriously?  ???

apzx

Re: Small Signal Diode Differences
« Reply #12 on: June 06, 2020, 07:52:43 AM »
> If you want to bias into Class AB, there is an optimum output current  which some authors call the "Oliver" condition https://www.hpl.hp.com/hpjournal/pdfs/IssuePDFs/1971-02.pdf page 11 describes this.

Seriously?  ???

Definitely Class Q.

JohnRoberts

Re: the "Oliver" condition
« Reply #13 on: June 06, 2020, 11:55:04 AM »
and what class of amplifier is Dr. Oliver referring to ??
class a/b ( I hope).

that HP note seems heavy on the math and pretty obtuse for a simple concept.

JR
Don't only half-ass tune your drums. Visit https://circularscience.com to hear what properly "cleared" drums sound like.

CJ

Re: Small Signal Diode Differences
« Reply #14 on: June 06, 2020, 01:11:14 PM »
build the circuit, you may find that the simulator and the actual circuit yield different results.

maybe that extra current for higher gain at high frequency is needed to charge e-b stray capacitance?
If I can't fix it, I can fix it so nobody else can!
Frank's Tube Page: www.mif.pg.gda.pl/homepages/frank/vs.html
Guitar Amps: http://bmamps.com/Tech_sch.html

apzx

Re: Small Signal Diode Differences
« Reply #15 on: June 07, 2020, 11:53:17 PM »
build the circuit, you may find that the simulator and the actual circuit yield different results.

maybe that extra current for higher gain at high frequency is needed to charge e-b stray capacitance?

I do intend to build a few different versions, but I'm going to be honest I don't expect to really here any differences. I'm sure I could probably measure the differences. At the end of the day most of the increases I got out of the circuit came from changing out the BC177 transistors for different ones. Though thinking on it some more I should go look at those again and see if there are correlations wrt to current in the stages or if it is just the transistor. And I did mention earlier (at least pretty sure I did) that I know sims are dangerous. I just thought if the sim is saying there is a difference, then there should be an explanation as to why. Whether or not that makes a difference in the real world is suspect until it is built and tested. Would I be particularly surprised if this thing oscillated and blew up it's output transistors? Not in the least ;)

Now, I knew about parasitic capacitances in BJTs, but what are you referring to when say charge the capacitance? I'm sorry I've sat and thought about what it is you're trying to say there and I've come up bumpkus :(

CJ

Re: Small Signal Diode Differences
« Reply #16 on: June 08, 2020, 01:08:51 AM »
as the frequency gets way high any stray capacitance will begin to steal energy.

but if the source is stiff then it will not matter.

it is a thought carried over from output transformer which have a lot of stray C which would normally cause resonant peaks when coupled with leakage inductance but in the the case of an output transformer, there is so much power that the stray capacitance has little effect on the flatness of the transformer as the energy consumed by the stray C is negligible compared to total  output power. probably BS but i want to sound like a genius so people will like me.  ;D
If I can't fix it, I can fix it so nobody else can!
Frank's Tube Page: www.mif.pg.gda.pl/homepages/frank/vs.html
Guitar Amps: http://bmamps.com/Tech_sch.html

apzx

Re: Small Signal Diode Differences
« Reply #17 on: June 08, 2020, 04:28:29 AM »
as the frequency gets way high any stray capacitance will begin to steal energy.

but if the source is stiff then it will not matter.

it is a thought carried over from output transformer which have a lot of stray C which would normally cause resonant peaks when coupled with leakage inductance but in the the case of an output transformer, there is so much power that the stray capacitance has little effect on the flatness of the transformer as the energy consumed by the stray C is negligible compared to total  output power. probably BS but i want to sound like a genius so people will like me.  ;D

Well I know next to nothing about transformers. That is a slight exaggeration, but really beyond mains transformers I don't really know a whole lot about them. I just like discrete opamps because I think they are absurd when there are ICs like the LM4562 or OPA1612 available. But because of the DOAs absurdity I love them :)

So, by opening this thread I have discovered so much more about the finer points behind opamp design in general. Regardless, of whether I end up using it or not. Honestly, I really want to use the M100 somewhere for something. Though, last I recall it does have something like 29.5mV of offset. I'm not sure how to get rid of that, and I would also really like to figure out a simple and elegant way to prevent a shorted output from frying the output transistors. So, much work to do! Though I suppose part of the offset could be removed by testing the input pair inside of the actual circuit and selecting them based on the lowest offset.

Edit - Actually the offset is largely determined by the bias current in the LTP. Grrrrr.
« Last Edit: June 08, 2020, 06:02:08 AM by apzx »

JohnRoberts

Re: Small Signal Diode Differences
« Reply #18 on: June 08, 2020, 09:21:51 AM »
as the frequency gets way high any stray capacitance will begin to steal energy.
in bipolar transistors the more significant capacitance is base  to collector aka "Miller capacitance" this acts like NF to roll off HF gain.
Quote
but if the source is stiff then it will not matter.

it is a thought carried over from output transformer which have a lot of stray C which would normally cause resonant peaks when coupled with leakage inductance but in the the case of an output transformer, there is so much power that the stray capacitance has little effect on the flatness of the transformer as the energy consumed by the stray C is negligible compared to total  output power. probably BS but i want to sound like a genius so people will like me.  ;D
In fact gate-source capacitance was a major issue for early MOSFET power amp designers using lateral MOSFETs for output device. Dropping MOSFETs in place of bipolar power devices in conventional designs would often run out of HF open loop gain when current starved driver stages could not supply enough current to slew the gate voltage fast enough at HF.  I suspect this accidental soft HF response contributed to some early adopter's attraction to the amps.

If you claim to sound better, it helps to sound different and the weak HF is audible.

JR
Don't only half-ass tune your drums. Visit https://circularscience.com to hear what properly "cleared" drums sound like.

apzx

Re: Small Signal Diode Differences
« Reply #19 on: June 08, 2020, 11:27:18 PM »
I'm debating if I should start another thread now because all my questions now are not pertaining to the diodes in this case. I'm pretty satisfied that the differences I was seeing are/were related to the differences that the diodes made to the current biasing the various stages. The main thing I was fighting after that was offset. I understand that without perfect matching and trimming there is no way that the offset is ever going to be incredible, but I'd like to think that the offset can be brought down into the sub 10mV range in the sims instead of nearly 100mV range.


 

Related Topics

  Subject / Started by Replies Last post
18 Replies
19259 Views
Last post March 02, 2006, 07:52:40 AM
by AMZ-FX
7 Replies
2280 Views
Last post March 31, 2006, 03:33:53 PM
by Matt Syson
16 Replies
23161 Views
Last post December 18, 2010, 10:57:46 AM
by electrochronic
16 Replies
2330 Views
Last post January 15, 2011, 10:40:19 PM
by dmlandrum