EHX DMM Questions

[POTL]

Hello everyone, I started to study the principle of operation of analog delay effects
I started to study the DMM scheme and it seemed to me that it looks very strange, in comparison with the BOSS scheme.
This forum has repeatedly helped me find answers to questions and I decided to ask a few more. 
1) I am confused by the bypass operation mode, in this circuit we have not a true bypass. The gain controller will work even when the pedal is in bypass mode.
Why is it so?
2) The first stage is an inverting amplifier, so the input impedance is R2 or 100K, which is below the recommended 500K-1M, is it a design error or is there any advantage in this?
3) What is this cascade made for? If you look at the BOSS DM-2 scheme, there are no additional amplification stages at all, only 3 stages of filtering on BJT, 2 filters at the input and output, and also the input buffer, which looks much more logical.
4) After the first stage of amplification, we have a non-inverting amplifier (IC1B) with a flat frequency response, but giving an additional gain of 6dB, why so many amplification steps?
5) after each delay chip, we have inverting amplifiers (IC4A & IC4B). Why are these amplifiers added to the circuit? They create additional noise and take up additional space on the board along with the components. Again, in the BOSS DM2 there are no additional amplifiers after the delay section and the pedal works fine without losing any loudness.
6) After the first delay chip (IC6), the outputs are R41 & R42 resistors, although in all other delay effects there is a trimpot for fine-tuning the balance. Why did EHX put resistors, because they do not guarantee a tonal balance setting?
7) The offset scheme looks standard, but after it there is no buffer, and this adjustment will affect the input impedance, in contrast to the BOSS pedals.
Is this also a design error?

In addition to questions to the scheme, there are a couple of additional questions.
8 ) Repetitions in the DMM sound brighter than in other delay effects, Memory Boy, DM2, Carbon Copy give blurry repeats.
What is the reason? This is due to the power supply of 15V input to the circuit or with filters in the audio part? This is interesting, since I want to collect a copy, but the MN3005 chips seem too expensive and I want to buy the MN3205. The problem is that the MN3205 only works from 9V, which means that if the brightness of the repeats depends on the increased power (15V), then my idea does not make sense.
What is the reason for the brightness and readability of repetitions? Filtration or increased power?
9) The NE570 chip remains a mystery to me. I understand that it works in compressor / expander mode, compresses the signal to the delay section and returns it to its original state after the delay section. But I do not understand what components affect, in Datasheet I did not find the information, where can I read it?
I will be grateful to your answers.

[PRR]

> IC1B... gain of 6dB

R6 R34 2-in mixer has gain of 0.5. So IC1_B is unity gain.

> after each delay chip, we have inverting amplifiers (IC4A & IC4B)

I don't see a IC4_B?

IC3_B IC5_B add gain. IC4_A corrects that.

> other delay effects there is a trimpot for fine-tuning the balance. Why did EHX put resistors

The very first BBD chips had poor balance across the two outputs. We trimmed. The trouble was just bad processing, which improved. As final trim is an eXpense, and chips got better, I'm sure they checked the chips they were getting and decided a trim was no longer essential. (It's not "tonal", it is clock joggle.)

> why so many amplification steps?..... Why are these amplifiers added to the circuit?

Sometimes it seems the designer got paid per-part, so padded-up his pay with extravagant use of parts.

Less cynically, parts are VERY much cheaper than boxes and knobs and mark-up, so nobody really cares anymore.

[POTL]

> IC1B... gain of 6dB

R6 R34 2-in mixer has gain of 0.5. So IC1_B is unity gain.

> after each delay chip, we have inverting amplifiers (IC4A & IC4B)

I don't see a IC4_B?

IC3_B IC5_B add gain. IC4_A corrects that.

> other delay effects there is a trimpot for fine-tuning the balance. Why did EHX put resistors

The very first BBD chips had poor balance across the two outputs. We trimmed. The trouble was just bad processing, which improved. As final trim is an eXpense, and chips got better, I'm sure they checked the chips they were getting and decided a trim was no longer essential. (It's not "tonal", it is clock joggle.)

> why so many amplification steps?..... Why are these amplifiers added to the circuit?

Sometimes it seems the designer got paid per-part, so padded-up his pay with extravagant use of parts.

Less cynically, parts are VERY much cheaper than boxes and knobs and mark-up, so nobody really cares anymore.

ic 4B in the lower right corner, in the BBD section between the two chips

Wait, ic 1B is a non-inverting amplifier, its gain is determined by R7 / R8
I just in case checked the connection of two resistors from two sources in the simulator and this does not affect the gain factor.

ic3b ic5b do add amplification, I did not take this into account, thanks)

Such a design of these filters is somewhat unusual for me, because you can recalculate the filter and not use the R23 R24 R28 R29 and C18 C22, this will give us zero-gain filtering, which means there will be no need for IC4A / B and accessories for them.
In theory, with the right selection of the filter, we do not need to hear the difference in sound, but we have to use fewer components, so the device can be made more compact.
Am I right or missing something?)

[PRR]

> I just in case checked the connection of two resistors from two sources in the simulator and this does not affect the gain factor.

If you drive *one* of the two inputs to the mixer, the mix loss is half, so the result is unity.



When driving both inputs with *different* signals, it is not simple.

Yes, unity-gain 2-pole filters are a thing. They can be fussy, or need extreme values, for sharp corners. Also before the internet we used whatever equations we could find, and the with-gain form is simpler and was once more common. (I don't think Lancaster's book even touches unity-gain?)

IC4_B (thanks) is a trim for the several-dB spread in BBD gain.

How compact do you need to be? With SMT parts you can probably make it smaller than the knobs.

[ElectricDruid]

1) I am confused by the bypass operation mode, in this circuit we have not a true bypass. The gain controller will work even when the pedal is in bypass mode.
Why is it so?

Non-true-bypass circuits were the normal thing until not so long ago. The holy grail at that point was "click-free switching", and things that provided a buffered bypass were seen as a benefit not a liability.

2) The first stage is an inverting amplifier, so the input impedance is R2 or 100K, which is below the recommended 500K-1M, is it a design error or is there any advantage in this?

No, there's no advantage in this. A higher input impedance would be better, but again, this is something that's only really become an important selling point in more recent designs. In the late-70s/early-80s no-one knew or cared, frankly. Except the engineers, and they weren't the ones buying pedals.

3) What is this cascade made for? If you look at the BOSS DM-2 scheme, there are no additional amplification stages at all, only 3 stages of filtering on BJT, 2 filters at the input and output, and also the input buffer, which looks much more logical.

3 stages or 2 filters on input and output sounds like a cascade to me. That doesn't sound so different. You've got to have a certain amount of gain and a certain amount of filtering - you choose how you want to implement that.

4) After the first stage of amplification, we have a non-inverting amplifier (IC1B) with a flat frequency response, but giving an additional gain of 6dB, why so many amplification steps?

The second stage is really a mixer for the input signal and the feedback signal. If the first stage had been used as a mixer as well, turning the Level up would have also boosted the Feedback - not what you want. It's not a conventional inverting mixer. Instead they've done a passive mixer into a non-inverting gain stage. You don't see this arrangement so much these days, and you can find articles online discussing the value of one approach over the other, but it's a valid choice.

5) after each delay chip, we have inverting amplifiers (IC4A & IC4B). Why are these amplifiers added to the circuit? They create additional noise and take up additional space on the board along with the components. Again, in the BOSS DM2 there are no additional amplifiers after the delay section and the pedal works fine without losing any loudness.

One of the English language's more horrible expressions is "There are lots of ways to skin a cat". E.g. you can achieve the same result in lots of different ways.
I doubt adding an extra op-amp in-between 4095-stage BBDs adds any noise that is significant. The BBDs will produce so much more noise than the op-amp that the op-amp's contribution will be entirely negligible.

6) After the first delay chip (IC6), the outputs are R41 & R42 resistors, although in all other delay effects there is a trimpot for fine-tuning the balance. Why did EHX put resistors, because they do not guarantee a tonal balance setting?

Good question. Maybe they decided that since the two chips were using the same clock it didn't matter and they could trim out the clock noise as far as possible on the output. Doesn't that work? Have you tried it? Why should the second chip care if the first chip provides a signal with clock noise, after all - the two are sampling at the same rate. It *should* be possible to simply balance the two signals at the eventual output. But I know that not everyone did it this way.
Ultimately, it's a compromise, like all designs. "Improve" it and add more parts, or keep it simple? Which do you choose? You can't do both.

7) The offset scheme looks standard, but after it there is no buffer, and this adjustment will affect the input impedance, in contrast to the BOSS pedals.
Is this also a design error?

Mostly, there are no "errors", only "decisions". Some of those decisions might look bad with time, but they probably made sense at the time. Occasionally, you can find a genuine error - I think there was one in one of the early ARP synth filters, which they finished up having to fix in a later revision.

8 ) Repetitions in the DMM sound brighter than in other delay effects, Memory Boy, DM2, Carbon Copy give blurry repeats.
What is the reason? This is due to the power supply of 15V input to the circuit or with filters in the audio part? This is interesting, since I want to collect a copy, but the MN3005 chips seem too expensive and I want to buy the MN3205. The problem is that the MN3205 only works from 9V, which means that if the brightness of the repeats depends on the increased power (15V), then my idea does not make sense.
What is the reason for the brightness and readability of repetitions? Filtration or increased power?

Almost certainly *not* the extra power. Why would that increase treble? The combination of filtering and companding is much more likely to be the cause. I haven't analysed the circuit closely, but do they do any pre-emphasis/de-emphasis too? They could boost the treble going in, so that when it is cut later by the filters, the cut is not so noticeable. This helps reduce noise/hiss where it is most objectionable and also makes the repeats seem more "present".

9) The NE570 chip remains a mystery to me. I understand that it works in compressor / expander mode, compresses the signal to the delay section and returns it to its original state after the delay section. But I do not understand what components affect, in Datasheet I did not find the information, where can I read it?

Thomas Henry's little book is the best source I know about the NE570 compander, but it's expensive for what you get:

http://www.lulu.com/shop/thomas-henry/making-music-with-the-ne570-compander/ebook/product-17459222.html

You'd find most of what's in the book on the web these days, with a bit of hunting.

[PRR]

> Mostly, there are no "errors", only "decisions".

Errors have happened.

It was an error to build the Comet jet-liner with squarish windows and a rivet hole in the corner. However this was not clear from the existing knowledge of aluminum fatigue, or revealed in extensive testing, and it took massive post-crashes testing to narrow down the sequence leading to blow-out and death.

A happier sequence on the Mosquito bomber. This was plywood; plywood they understood. Prototypes are expected to stand 120% of design load. The first proto's wing fell off at 60%, and it is suggested they expected this. They beefed it up in stages to 116% of design load. The Mosquito was an excellent warplane in part because it was very light for the work it did. (OVER-building is a different kind of error-- some German WWI planes were built and flew like trucks, lighter Gnomes flew rings around them.)

[POTL]

> I just in case checked the connection of two resistors from two sources in the simulator and this does not affect the gain factor.

If you drive *one* of the two inputs to the mixer, the mix loss is half, so the result is unity.



When driving both inputs with *different* signals, it is not simple.

Yes, unity-gain 2-pole filters are a thing. They can be fussy, or need extreme values, for sharp corners. Also before the internet we used whatever equations we could find, and the with-gain form is simpler and was once more common. (I don't think Lancaster's book even touches unity-gain?)

IC4_B (thanks) is a trim for the several-dB spread in BBD gain.

How compact do you need to be? With SMT parts you can probably make it smaller than the knobs.

I did the simulation again capturing a little bit of most of the circuit, you're right both resistors affect the gain, I made a mistake, thanks for your answer)
I am interested in diley in the case of 125B, I use mostly SMD, but in addition to the compactness I'm interested in the same circuit reduction, vactically I think about how to collect a full clone of DMM and in parallel to collect the second diley making adjustments to it and in practice to understand how important these or other stages of the scheme and how you can change it without losing sound.
Thanks.

[POTL]

1) I expected it to be echoes of the past)
2) Another classic point
3) ОК
4) I already knew I was wrong.)
Now looked again at the diagram of the BOSS DM-2 and adjust the Intensity(Feedback) there are organised differently, the feedback is after the filter from the compressor, it is likely that the reason that repeats sound dark.
5) Indeed, it is worth checking this point in practice.
6) Yes, I was prompted on another forum that it makes no sense to rebuild both chips, although again you can check in practice whether there is any difference between these methods.
7) ОК
8 ) Unfortunately I have not had any experience with BBD, I can only judge from the experience with distortion effects, when the increase in power the sound becomes better and more voluminous, I read that ANALOGMAN claims that more power improves the performance of his chorus.
9) I will try to find a cheaper source of knowledge 

Thank you very much for the detailed answers, you helped me to understand this scheme better.

[DrAlx]

Inverting between BBD stages can lower the clock noise.  This was discussed on another thread which I can't seem to find.
I can't recall if it was the DMM, but I do remember a chain of at least 4 BBDs with inverting op-amp stages in between.
The idea is that the inverter flips the clock glitch from the previous stage so that the clock glitch from the next stage tends to cancel it.
It makes sense if you take a look at what a typical clock glitch waveform looks like on a scope.

[POTL]

Inverting between BBD stages can lower the clock noise.  This was discussed on another thread which I can't seem to find.
I can't recall if it was the DMM, but I do remember a chain of at least 4 BBDs with inverting op-amp stages in between.
The idea is that the inverter flips the clock glitch from the previous stage so that the clock glitch from the next stage tends to cancel it.
It makes sense if you take a look at what a typical clock glitch waveform looks like on a scope.

interesting idea

[ElectricDruid]

> Mostly, there are no "errors", only "decisions".

Errors have happened.

They surely have. I mentioned the ARP filter.

There are some which it isn't clear which category they're in. One example from the Synth world is the Roland TB-303's famous filter. It has a somewhat-bizarre response and a very nasal resonance which might well have been an error, but which has now become a distinctive feature and is widely copied/cloned and imitated.
In a similar way, I expect Bob Moog would have been happy to get rid of the distortion in his ladder filter design on the day he designed it. After he became famous in part because of the distinctive sound that distortion gave the filter he invented, I doubt would be so keen to remove it. Several later Moog synths kept certain elements that would have been considered "obsolete" or "superseded" and it's likely they did that because by then they'd realised those elements were a key part of the "Moog sound".

A happier sequence on the Mosquito bomber. This was plywood; plywood they understood. Prototypes are expected to stand 120% of design load. The first proto's wing fell off at 60%, and it is suggested they expected this. They beefed it up in stages to 116% of design load. The Mosquito was an excellent warplane in part because it was very light for the work it did. (OVER-building is a different kind of error-- some German WWI planes were built and flew like trucks, lighter Gnomes flew rings around them.)

The error I know about with the Mosquito design was using a casein-based glue and then flying the plane in the tropics. They started to delaminate and literally fall to pieces until they changed the glue formulation.

[j_flanders]

2) The first stage is an inverting amplifier, so the input impedance is R2 or 100K,

2) Another classic point

I was hoping for a more interesting and compelling reason as well...
Something along the lines of :

- It loads down a passive pickup (lower resonance peak, lower Q) making it sound less bright and so the dull repeats no longer sound so dull in comparison to the dry signal and will blend in better.

- It provides a nice interaction between the guitar's volume knob and the gain of the first opamp. A quick Google search hower doesn't show anyone raving about this 'feature'.

Surely there must have been a good reason to not go for a non-inverting setup there. I know the general rule is to "always invert, unless you just can't". But this, to me, looks like a very typical you can't or don't want to scenario'.

[PRR]

> Surely there must have been a good reason

Sometimes choices are arbitrary.

Here there may be no "wrong" choice. It works. Either way.

In commercial warranty design there is one "good reason" to do it this way: the idiot users can put in hundreds of volts (from a speaker jack, or mis-wired gear) and nothing blows-up. While there are other ways to make a sensitive input blow-up resistant, this is easy.

When designing a LOW-hiss amplifier, series resistance (inverting config) is a strong no-no. However e-guitar is on the stronger range of what is really low-hiss design. A marginal case. Balance your choices and flip a coin.

The 100K must be a compromise between loading-down the top of the guitar and the added hiss. 10K is clearly too brutal. 1Meg is awful hissy for guitar chain.

An added commercial advantage is: when you plug-in to the input, but don't have a source (guitar) on the other end of the cord, the hi-Z non-inverting form will buzz loud, and powerfully if you tap the tip with your finger. This is most likely to happen in the show-room! The inverting form goes to unity gain or less, buzz is less, but finger-buzz (a useful diagnostic) still happens on the high overtones.