Why not use a single trimmer across the base in a Rangemaster?

[soggybag]

I built a Rangemaster recently and thought, why not use a single trimmer in place of the 470K and 47K resisters that go across the base of the Ge?

Seems you could dial in the same voltage with a 500K or 1M pot. Seems this this would be easy and you could dial in the voltage to set the bias.

[R.G.]

Good idea... if it wasn't using a pot to conduct both DC and audio signal. It'll be scratchy like crazy if you move it while you listen.

Even if you plan to use it for a set-and-forget control, you don't need all that adjustment range. The usable range of voltages on the base of a rangemaster clone is quite small. I haven't measured or calculated it based on the variables of possible transistors and leakage, but it's probably way less than a volt. Having a base adjustment outside that range is a hindrance, because the usable bit of rotation is all squeezed into a tiny portion of the pot rotation.

Better to put the pot in the middle of two other resistors. Say, use a 10K pot with a 470K and a 39K at either end. That way the whole pot rotation is going to be more usable for adjustments.

[soggybag]

I was thinking to use this just to set the bias. I noticed that the Java Boost has two trimmers, which made me think, why not use one?

[R.G.]

I would certainly use one trimmer instead of two!

Edit: I realized I should explain myself.
Pro designers think of trimmers as the spawn of the devil. This is the hierarchy of "goodness" in designs:

1. The best designs work first time, every time, with any mix of parts which are in specification.
2. Next best, designs for which there are some critical components which must be selected, but otherwise just work.
3. Next, designs which must have some measurement, then a selected part.
4. Next, designs which must use one trimmer adjustment to work right.
5. Worst, designs which must use more than one trimmer.

More than two trimmers means either (a) the designer doesn't understand the circuit well enough to do better, or (b) parts are simply not available to do the job. I once saw one of my classmates in our circuits-design class put in pots for *all* the resistors in a two transistor circuit and tweak it in from there. His final tweaking worked perfectly to the requirements of the assignment, and the instructor gave him a D, saying that since it did work he couldn't in good conscience give him an F for not understanding what he was doing wrong.

Pots are evil. Pots are mechanical devices that have a direct effect on the perfect function of your circuit that **users!!** can touch and change. They wear out, are hard to find, have miserable tolerances, drift, are noisy, get intermittent; worse, they HAVE to be adjusted on the final assembly line by a "tech" who may not have finished high school, let alone the training program to do the adjustments right; worse than that, they WILL be misadjusted at random by users who make that "tech" look like Albert Einstein by comparison.

Wait... must... breathe... slow... ly...  ah... centering... better.

Sorry. Let's just say that trimmers produce issues in any design which is ever getting off the workbench. 

[Taylor]

Good idea... if it wasn't using a pot to conduct both DC and audio signal. It'll be scratchy like crazy if you move it while you listen.

Just write "Crackle OK" on it.  This is how the pros do it, RG, keep up. 

[R.G.]

Just write "Crackle OK" on it.  This is how the pros do it, RG, keep up. 

Doh!! Of course!!  Where is all the training from my Zen masters? If you can't fix it, feature it! 

That was the first rule of the Zen marketing masters.

I will try harder. I'm pretty sure I saw a grasshopper here somewhere... 

[Gus]

When I build a Si RM type circuit I use a fixed resistor and pot in series for the "top" resistor in the bias section. 

[frank_p]

Where do sockets goes in the hierarchy of "goodness" in design ?  I've heard several times, by Gus, that: "trim pots and sockets is not design".

1. The best designs work first time, every time, with any mix of parts which are in specification.
2. Next best, designs for which there are some critical components which must be selected, but otherwise just work.
3. Next, designs which must have some measurement, then a selected part.
4. Next, designs which must use one trimmer adjustment to work right.
5. Worst, designs which must use more than one trimmer.

I suppose they must fit in a similar "hierarchy".

I may be wrong but: trim pots and sockets are still tools for learning in the lab: the scientific process of changing one variable at a time in the circuit and measuring the results is a good way of knowing how what you are studying reacts.  Once it's boxed and goes in the world, it's an other thing: no?  (Well, as far as I know you don't put a breadboard in an Hammond enclosure...)

Simple example: yesterday I was working on a modified version of the transistor tester that is on GEO.  The trick of measuring the voltage on the resistor at the leads of the collector resistance did not work.  I had an ampmeter at he base and an other at the collector and a volmeter across the collector load.  So, I've putted a trimpot and a fixed resistor at the collector to set it at the specified values so to have the gain on the screen of the voltmeter.  The problem is that the gains obtained by dividing the currents (185) and the one by measuring the voltage (160) were different.
Hypothesis:
- the DMMs internal resistance is disturbing the bias setup;
- The DMMs are not precise an off;
- The calibration of the DMM is offset;
- Changing the DMMs scales is changing the internal resistances;
- there is leakage;
- etc.
- etc.

By re-setting the trimpot it was more easy to make the changes so that the two processes of measuring the gain match each other.  And, If I change the three DMMs to other ones I can trim to compensate for the change of parameters in the measuring equipment.

By the way:
What is the class note for putting trimpots in sockets.  G : banned from school ?

Gus just posted too fast for me...

[Gus]

I should have posted it is an external full size pot.  I can design for the operating point I like: however I like to adjust bias for different "tones"  In a RM type I like a range from about 7. something VDC to under 5VDC.

The "best" way to do this might be to use an opamp or EF and noiseless bias.  The opamp would be a buffer with a adjustable voltage divider input.  The out would be connected to a 100 ohm resistor and a cap to ground.  The input resistor would connect to the 100 ohm cap node and then to the base.  This would keep the input resistance the same with just a bias shift.  I also adjust bias with other active circuitry.

[JDoyle]

Pro designers think of trimmers as the spawn of the devil.

Unfortunately, they have to make a deal with the devil in almost every case, especially in audio.

Using JFETs? You will need a trimmer. Period. Even for parts from the same batch. Especially when you see that sorting for the proper characteristics will end up with enough parts out of spec that you could buy 5 trimmers for every 10 parts tested.
Discrete differential input, DC Coupled circuit? You will need a trimmer to zero the offset.
Opamp circuit requiring DC amplification (servos)? You will need a trimmer to zero the offset of the OP AMP - or one setting the bias voltage on the servo to cancel the internal offset of the op amp - and all the transistors in THAT are as identical as you can get!
Biasing the output of a power amp? Variation in transistor gain/Vbe require a trimmer to set the proper bias. Every example I have seen using a Vbe multiplier uses a trimmer. From Aiwa to Krell and Levinson.

I think it should be noted that the 'pro designer' outlook outlined here is that influenced by the requirement to reduce costs at every point in the manufacturing process - a belief hammered into EE students at every step and adjusting a trimmer takes time=costs money - and not by any 'quality of design' metric associated with the final product as used by the consumer; especially if the quality under scrutiny is whether or not something sounds good.

In my opinion, even in mass manufacture situations, if the circuit requires a trimmer to adjust to the optimal conditions, and that trimmer is a 10-turn closed frame type, and not an open frame, single turn carbon type, the design and the final product aren't a priori deficient or born out of lesser/'amateur' (as opposed to 'pro') design skills.

There are situations where the bottom line isn't first on the list of importance, and in my opinion audio is most assuredly one of those.

Of course, if it is possible to design out a trimmer without any effect on the final circuit then by all means this is the proper design choice, but if one can't, one is not a lesser designer.

Regards,

Jay Doyle

[R.G.]

Unfortunately, they have to make a deal with the devil in almost every case, especially in audio.

Yep. As I noted, I'd certainly use one trimmer if the alternative was two. And the description was tongue in cheek, just to be clear.

Did you notice the caveat? "(b) parts are simply not available to do the job"
Sometimes there is no other way. But there can be no question that any design would be better if a way could be found to do the same thing without a trimmer. Trimpots are what you do when you can't think of anything else to do.

Using JFETs? You will need a trimmer. Period. Even for parts from the same batch. Especially when you see that sorting for the proper characteristics will end up with enough parts out of spec that you could buy 5 trimmers for every 10 parts tested.

Yep. This all by itself is sufficient to explain why JFETs, with all their other advantages, lost out to bipolars in the electronics world. If you have to use a trimmer per JFET, and your competitor can do the same thing without trimmers by not using JFETs, you go broke on service and manufacturing costs.

Discrete differential input, DC Coupled circuit? You will need a trimmer to zero the offset.
Opamp circuit requiring DC amplification (servos)? You will need a trimmer to zero the offset of the OP AMP - or one setting the bias voltage on the servo to cancel the internal offset of the op amp - and all the transistors in THAT are as identical as you can get!
Biasing the output of a power amp? Variation in transistor gain/Vbe require a trimmer to set the proper bias. Every example I have seen using a Vbe multiplier uses a trimmer. From Aiwa to Krell and Levinson.

Yep. That's what you do if you can't do anything else. But did you ever hear of a commutating auto-zeroing amplifier? The specific intent of those is to get rid of the zeroing procedure on DC amps. And one of the touted virtues of any opamp claiming to be "precision" is how low it's input offset is. The thrust of a lot of IC engineering is to get rid of those mismatches. Many of the remaining offsets in opamps are from internal thermal offsets, not mismatches in the devices themselves. And power amps? Sure, we still need to trim the bias on those - until we get to to either reliable auto-biasing circuits (one of which I designed for a tube amp not all that long ago) or until we get to class D, which needs no biasing at all. I don't think any of this is at odds with the idea that trimmers are not what you'd do if you had a choice. And by using more trimmers, worse yet interactive trimmers, things get seriously out of hand.

I think it should be noted that the 'pro designer' outlook outlined here is that influenced by the requirement to reduce costs at every point in the manufacturing process - a belief hammered into EE students at every step and adjusting a trimmer takes time=costs money - and not by any 'quality of design' metric associated with the final product as used by the consumer; especially if the quality under scrutiny is whether or not something sounds good.

I had to listen to enough lectures on "how do the Japanese do quality designs", "six sigma design criteria", and product quality lectures that I'm pretty well versed in quality issues, at least as the commercial world sees it. Quality is conformance to specification. Period. Otherwise, it's just a taste test, and you may get as many results as you have tasters. "Sounds good" may be a quality criterion, but how in the world are you going to figure out that 99.9996% of your products meet the criterion?

Besides, "sounds good" is not the issue. The issue is goodness or badness of trimmers on their own merit. If you have two designs, each of which is otherwise the same, with same goodness of sound, same cost of parts except for the trimmer(s), same manufacturing cost except for the trimmer(s), which is better? I think it's the one which the uninformed user/service guy cannot muck up by twisting that trimmer to see if it gets even better, the one which does not have trimmer failures, the one which does not get returned as defective when all that happened was that shipping vibrations moved the trimmer, and the one which when the end-of-assembly guy is behind  and needs to hurry up, the trimmer does not get set.

The idea that we can make a design better in some way which has nothing to do with the other results is not antithetical to end design quality. Sure you have to achieve good results with your designs; in the audio world, they have to sound good, all of them. That's an entry criteria. After that, the quality of the innards of design rely on who can get that same result with fewer (and perhaps better!) parts, lower cost and faster time to market.

In my opinion, even in mass manufacture situations, if the circuit requires a trimmer to adjust to the optimal conditions, and that trimmer is a 10-turn closed frame type, and not an open frame, single turn carbon type, the design and the final product aren't a priori deficient or born out of lesser/'amateur' (as opposed to 'pro') design skills.

I guess I agree with that, but I'd say it differently. It's not necessarily an amateur design job just because there's a trimmer in it. As I've already said, sure, you do what you have to get the design to meet specification. As one of my personal heroes, Albert Einstein said, everything should be as simple as possible, but no simpler. I would never argue that trimmers will never be needed; but I think you'd agree that if there is a way to get the same results without them, that's a superior design job - right?

There are situations where the bottom line isn't first on the list of importance, and in my opinion audio is most assuredly one of those.
Of course, if it is possible to design out a trimmer without any effect on the final circuit then by all means this is the proper design choice, but if one can't, one is not a lesser designer.

I believe that this is a restating of what I just said; and that you're in what we used to call violent agreement with me.

More to the point, do you disagree with my heirarchy of designs, given that any design must get to the same eventual conformance to requirements?

[mac]

soggybag,
if it is for your personal use, a single trimmer is fine, but what about noise across the trimmer? RM is known to add noise.
If trimmers are noisy then replace the 68k or the 3k9, not the whole 470k + 68k.

BTW, if the transistor you are going to use have the right gain, near 70 and leaks less than 0.1ma it should bias near 6.6v -7.0v.
Mine has a 4k7 and bias at 7.0v. Transistor is a Matsushita 2SA102 hfe=70, leakage=150ma.
Have you tested your transistor or tried the original RM bias?

mac

[darron]

just a confirmation that i've done it in a very similar circuit, so it can be done.

a normal trimmer isn't much use, as R.G. points out about the adjustment range.

i've used a single 500k multi-turn, no good results from a 1m though.. if you don't mind pushing the parts count up a tiny bit then i'd probably only adjust one of the values as the DC biasing point seems like a very sensitive point to be adding noise while the two values 'fight it out' to make the voltage you need.... aren't i imaginative with my understanding of electronics? hehe

[R.G.]

... the DC biasing point seems like a very sensitive point to be adding noise while the two values 'fight it out' to make the voltage you need.... aren't i imaginative with my understanding of electronics? hehe

Actually, that's a pretty decent description! 

[DougH]

For building one-offs for myself, I avoid trimmers if at all possible. Most of the time there's no need for them. If it's enough of a "feature" (as in Gus's bias pot) then I use a normal pot on the control panel and stick a knob on it. Otherwise, I try to avoid as many mechanical contraptions as possible in an electronic design, for all the aforementioned reasons.

If I was manufacturing I would do everything I could to design around device inconsistencies and avoid things like trimmers and jumpers. I bought a vibe a couple weeks ago and tried it before I left the store, and it just didn't sound right. So I swapped it with the one in the demo area because that one sounded good. I'm 99.9% sure it just required a bias trimmer adjustment. But when I got home I took a look and was glad I didn't bring home the "bad one" since it required more disassembly than I thought just to get to the trimmer. A trimmer-less design or at least custom fixed resistors would have been much preferable. But I do see a lot of trimmers used in audio stuff.

[mac]

i've used a single 500k multi-turn, no good results from a 1m though..

9v/1M = 0.009ma is to low for a low leakage Ge having a gain of 50-100 since base current is (9-7)v/10k/70 = 0.0028ma.
Base current should be tiny compared to the current across the base resistors.
A leaky Ge might work with a 1M but leakage could send the collector down.

mac

[JDoyle]

Trimpots are what you do when you can't think of anything else to do.

I was responding in a pure audio outlook so in the case of the quote above I would add: '...or if what you do think of doesn't sound as good as using a trimmer.'

Yep. That's what you do if you can't do anything else. But did you ever hear of a commutating auto-zeroing amplifier? The specific intent of those is to get rid of the zeroing procedure on DC amps. And one of the touted virtues of any op amp claiming to be "precision" is how low it's input offset is. The thrust of a lot of IC engineering is to get rid of those mismatches. Many of the remaining offsets in op amps are from internal thermal offsets, not mismatches in the devices themselves. And power amps? Sure, we still need to trim the bias on those - until we get to to either reliable auto-biasing circuits (one of which I designed for a tube amp not all that long ago) or until we get to class D, which needs no biasing at all. I don't think any of this is at odds with the idea that trimmers are not what you'd do if you had a choice. And by using more trimmers, worse yet interactive trimmers, things get seriously out of hand.

Ignoring Class D, which in my opinion, though quite amazing in it's efficiency, isn't what I myself would consider 'high quality' (to avoid the use of 'hifi' so I don't get tainted with the crazy brush) audio design - a downside of both the auto-zeroing amplifier, and auto-biasing circuits is that you now have a much larger component count, as much as doubling the circuit size, leaving, simply, more stuff to go wrong. And in the case of servos, auto-zeroing, etc. if (we both know that should read 'when' in at least a percentage of the cases) that circuitry goes wrong, lets say in a power amp, a lot more down the line could or will get burned up - including the speakers and output transistors. In the case of a Vbe multiplier, if we are smart and protect the circuit, at the cost of a single resistor, from the trimmer becoming open circuit, nothing is damaged, the amp may still work if the resistor is chosen correctly, and the fix is a cost of a new trimpot, not a set of power transistors, whatever speakers were attached, etc. Now if I was a consumer, and had both of those situations occur, which amp would I consider a higher quality design? That which doesn't need a trim but risks catastrophe, or that which has a trim?

Besides, "sounds good" is not the issue. The issue is goodness or badness of trimmers on their own merit.

I am assuming that we are talking about audio designs, I'm guessing the original poster and everyone on this board will be playing through the circuits we discuss, in which case there simply isn't any way to separate out the sound quality, or at least ignore it altogether to concentrate on the pure goodness/badness of trimmers. To do so is an interesting intellectual exercise in design philosophy, but in my opinion ignores the reality of audio. If we are talking about a toaster, sure all of your statements apply in full, but we are talking about something entirely different - the pursuit of musical inspiration, through electronic circuits, on this board, and in almost every situation, no matter how well designed, no matter how tight tolerance of parts one has, there is almost always going to be a need to tweak to get it 'just right'.

If you have two designs, each of which is otherwise the same, with same goodness of sound, same cost of parts except for the trimmer(s), same manufacturing cost except for the trimmer(s), which is better? I think it's the one which the uninformed user/service guy cannot muck up by twisting that trimmer to see if it gets even better, the one which does not have trimmer failures, the one which does not get returned as defective when all that happened was that shipping vibrations moved the trimmer, and the one which when the end-of-assembly guy is behind  and needs to hurry up, the trimmer does not get set.

Of course the bulletproof one is better, but seeing as men much more talented than myself, men who have spent there lives studying and building audio equipment, in the end still a trimmer in their designs, shows me that in audio, Goldilocks resides in the ether, and one must find a way to tweak the circuit to get it 'just right'. For example, Douglas Self in his Amplifier Handbook has a 'blameless' Class B, with pretty much every source of distortion possible minimized to the greatest extent. On paper, to specification, I can't think of a better amplifier, but right there, in the output stage sits a trimmer. Does this mean that he was unable to come up with a more 'quality' design that eliminated that one trimmer? And is that trimmer, by it's very nature of being in the circuit, 'bad' or 'evil', making Self a lesser designer?

I see your point, I just think it impossible to judge the merits of a circuit meant for audio without taking into account the actual sound. If one doesn't, one ignores the entire issue, in my opinion. If one does, then one must decide whether the problems you mention are to great to make the design manufacture-able or marketable - but even if the answer is no, that doesn't automatically consign it to the dustbin of 'poor design' just that it wasn't easy to make a buck off of it.

As one of my personal heroes, Albert Einstein said, everything should be as simple as possible, but no simpler. I would never argue that trimmers will never be needed; but I think you'd agree that if there is a way to get the same results without them, that's a superior design job - right?

I think this is debateable on both points - is the servo biased amp more simple than one with a Vbe multiplier with a trim? I can't see how it would be requiring at least two large and poor quality electrolytics, an op amp, and several resistors, plus rail bypassing for the op amp, the op amp power connections vs. a trim, at most two resistors, and a transistor, with only two points connecting to the actual circuit. Plus like I said above, if one fails, the amp at worst it sounds like crap, the other fails and the whole system could go down with it.

More to the point, do you disagree with my heirarchy of designs, given that any design must get to the same eventual conformance to requirements?

I do agree with you - for a toaster. But I disagree that one can ignore whether or not the circuit sounds good, when that is the reason and essence of why the circuit was even conceived in the first place. And to hold the same design criteria for a piece of audio equipment, as you would for a toaster, just isn't in my mind a very rational outlook.

It isn't that I disagree with your reasoning or core points, I just think that it is a mistake to apply one single idea of what constitutes 'good design' across every sphere of circuit design and not take into account the purpose of the circuit's existence in the first place.

I guess what I'm saying, is that I don't think in most cases involving audio, that it is possible to design out the trims necessary to get the circuit's 'just right' without it being detrimental to the whole design, or the manufacturing process, even with BJTs the variation will cause the need for subtle tweaks, or you get the 'fuzz face' effect, where some sound great, and like they were supposed to, but others don't. And by removing the trim, you removed the ability to make them all good. Which in a way was what you were saying...;-)

[darron]

back on the actual circuit, would putting a small value (< 1K ohm) in series with the signal, after bias trimmer entry, help to lightly regulate the noise a little introduced by such a primitive bias method (which used to be carbon resistors anyway)? it could double as transistor protection from too hot of a signal (: