Measuring the impedance at the FX loop

[Stephen1966]

Okay, a bit of background (better late than never!)

There were a couple of reasons why I wanted to explore the problem of what MrD meant when he stamped the FX loop of his 124 chassis with '2.5 VOLTS RMS - .5M OHMS.' Martin's theory that 2.5Vrms represented both the limit of a clean, unclipped signal, and (inferring) a safe operating level for 9V effects, along with the expected load of effects at 500k is fairly compelling for me. A second reason, was to determine the maximum voltage in this part of the circuit because, having tried a passive prototype of a loop, I am unhappy with the results and I am now considering an active FX loop. However, in order to design the active circuit with a realistic input level, I needed to know the max voltages we are looking at in my design. I will come back to the passive prototype shortly, but first @sluckey I went back to the amp and ran the tests again using a 37mV input signal.

Once again, it was the clean channel only but following the Fender process, I turned the bass, mid and treble all the way up and set the probes to 10x with AC coupling. The 37mV sine is noisier than the 1V, and this gave me some erratic Vpp readings.

Here are the traces with the volume and master at full with a 37mV input:

Vol full and Mast full.jpg

For a 37.29mV(rms) input signal (blue trace), I measured 150mV(p-p). Using the formula I posted earlier, it should have been 105.47mVpp. However, I'm not going to let a few stray millivolts break up the band so reading the signal at the wiper on the master pot, I read 17.83Vrms and 52.16Vpp. The caulculated Vpp is much closer this time at 50.43V so Steve is right, it does clean up the maths. This, is pretty much the same result I got before with the Volume and Master at full with a 1V input signal (actually closer to 900mV) so at the top end of the dials for M and V, it doesn't result in any big changes but the maths, and the signal are cleaner... tighter.

I think this next trace shows how it cleans up the signal. This is with the volume at midnight and the master at 100%

Vol half and Mast full.jpg

What I am really interested in at the end of the day is the way the signal looks at something like a playing level. This is now with the V and M pots both turned up to about midnight - which since I'm using j-taper pots there, is more or less 30% of max.

Playing level.jpg

The signal now looks quite symmetrical and clean and it shows something of the dynamic (touch sensitive) nature of these amps that when you don't drive the input, you can can get a very clean signal. You will notice it isn't exactly the inverse phase of the input signal; they aren't exactly 180° apart. If anyone could enlighten me what causes there may be for this slightly "out of phase" timing I would be grateful, and it might go some way towards improving or reducing some of the distortion that happens when the input level is much higher.

All in all though, an eye-opening experiment. A big thanks to sluckey (hope I got your name right) for keeping me sane

[Stephen1966]

So now to the passive prototype I have been working on. Some observations why it works, and why (ulitmately) I am unhappy with my design. The inspiration for this came from Merlin Blencowe's excellent Designing Tube Preamps for Guitar and Bass. To be specific, his design 'Fig. 12.10: Practical parallel effects loop as used in the Engl Fireball.' Page 220. Now, if you come across this post, Merlin, don't be offended. You might be a Burnley lad, and if that's case, you should know I am a Bacup lad: it's just who we are!

This design works. Merlin is conscientious with his published designs and he wouldn't have kept this work in print if it didn't. However, mine was not its intended use, hence the next revision, but I am happy to report here some features of this design, and a bit more detail about how it works.

Here is my design schematic:

SKYLINER passive FX loop 2.6.jpg

Here is the layout:

Skyliner 2.6 FX loop LAYOUT.jpg

And a photo:

IMG_20220715_164427.jpg

At the heart of this design, you might say, is the parallel circuit dual gang Mixer pot, wired antiphase so that as the dry signal increases, the wet signal decreases in exact proportion. This design does that. However, as I will show in the next post (the upload limits on a single post mean I have to split this over several posts) it is not a good idea to use an audio taper dual gang pot in this position. True, audio taper more accurately reflects the way we hear the (logarithimic) increase in amplitude of a signal, but it produces uneven results when used in this application. At the time Merlin's book went to press, I don't think Merlin had had the opportunity to get hold of a dual gang log pot, his design references a linear pot. On paper, it sounds like the audio taper pot was going to be more "lifelike," but in practice, it probably doesn't work as well as a linear pot would.

Next post...

[martin manning]

The Vrms/Vp relationship of Sqrt(2)/2 only applies to a pure undistorted sine wave. Your scope is making an rms calculation for whatever waveform it has (distorted and asymmetrical as it may be) based on some set of sampled points, and in that process it must have some way of determining a zero reference. Given the analog nature of the times, I have to believe Mr D was speaking in reference to pure sine waves.

The phase of the signal at the preamp out is determined by the number of inverting stages and any RC filters in the signal path. A basic low-pass filter shifts phase somewhere between 0 and -90 degrees depending on the frequency of the signal wrt the corner point of the filter. Likewise, a basic high-pass will shift the phase between +90 and 0 degrees. Of course there are several such filters along the path from the input to the preamp output, and if you change your test frequency you will see a different phase shift.

[Stephen1966]

Looking at the action of the Mixer pot in more detail here I tested it first in parallel mode, that is, mixing the dry and the wet signals together, in three basic settings: full CCW (max dry/min wet), at half/midnight (wet and dry equally mixed) and, full CW (min dry/max wet). First then, lets see the how the signal looks at the half/midnight position:

Mix at half - parallel.jpg

This is with the same 37mV/1kHz input signal. This time, the yellow (reference) trace is the signal coming of the Master, wiper and the violet trace is coming off the Return jack tip.

If we compare these with the signals we see when the Mixer pot is turned CW and CCW we see a significant drop in amplitude.

Fully CW:

Mix at full CW - parallel.jpg

Fully CCW:

Mix at full CCW - Parallel.jpg

Now, I didn't make a note of the difference between these peaks but I think it's clear to see that when the mixer is turned either fully CW or CCW there is an obvious drop in amplitude and you can read that in the FFT trace underneath the signal traces. I would venture to say, this is a property of the taper of the pots. Both the CW and CCW levels look the same but they are out of kilter with the halfway setting and if you are looking for a more consistent level at all positions, you would be better of going with the linear pots.


Next post...

[pdf64]

This can be set for frequency or phase response plots https://www.guitarscience.net/tsc/dumbl ... V_pot=LogA

[Stephen1966]

When it comes to the circuit in Series, with the signal going only along the wet path, there was perhaps an even more idiosyncratic feature, one that I wasn't expecting, but when you follow the logic of the circuit, it makes perfect sense:

Here is the wet signal with the mixer pot fully CCW:

Mix at full CCW - series.jpg

Here with the mixer at midnight:

Mix at half - series.jpg

And here, with the mixer fully CW:

Mix at full CW - series.jpg

I think you can see more clearly from the FFT traces the way that the mixer pot acts as an attenuator in series mode. Increasing the signal in amplitude as you turn it CW; and attenuating it as you turn it CCW. If you employ this circuit in Merlin's design it may be worth noting this. Merlin doesn't mention it but if you want something like an unattenuated signal similar to what you get in the parallel mode with the mixer at the half way position, you have to turn the mixer all the way CW in series mode. You can't have the maximum signal in series and parallel with the mixer pot in the same position. In Merlin's practical loop design, in series, it will function like a return pot in other active designs. It simply adds another layer of complexity to its operation when you want to switch between series and parallel, though I'm sure there is a sweet spot somewhere on the CW side of the dial where the series and parallel levels are more or less the same and where switching between them wouldn't require any adjustment. Without relay/footswitch activation or switching between these two realms, it's likely a moot point, since you would be simply setting and forgetting it at the sound check. Even so, it's a consideration you may need to take account of.

This wasn't the reason I started to move away from this design.

Next post...

[Stephen1966]

Attenuation!

This was the deal breaker for this design. It comes when you switch between bypassing the circuit and engaging it. It's there you see the difference in the loading it puts on the signal.

Here is it in Engaged mode:

Engage.jpg

And here it is when the loop is bypassed:

Bypass.jpg

In a low impedance, CF driven active loop, I am sure this wouldn't be such a problem but it's significant and really unusable in this, my passive design. As I said at the start, I wasn't using it for it's intended design so there's no-one to blame but myself here. The attenuation can be compensated for, of course, with complimentary volume and master levels but that also lifts the sound floor. So no! A good parallel circuit, with an interesting side feature when it's in series mode but not much good in a passive design. The problem of achieving unity gain with an active loop in engaged/bypassed and series/parallel modes can all be solved with an additional send pot followed by the dual gang mixer on the return. Again, I think you would have to play with the mixer for a balance of signal strength between series and parallel modes but the send pot could compensate for that attenuation I am seeing in the engaged passive circuit.

It's onward and upward with a new active loop design for me now. Along the way, this experiment and it's failure in this specific application, has taught me a number of things. I'll say it again, the design works, and I'm sure if you consider it in your loop designs (and pay heed to its idiosyncrasies) ultimately, you won't be disappointed.

[Stephen1966]

This can be set for frequency or phase response plots https://www.guitarscience.net/tsc/dumbl ... V_pot=LogA

Excellent! Thank you. It's been ages since I looked up the TSC. I don't remember seeing the Dumble sim before.

[Stephen1966]

The Vrms/Vp relationship of Sqrt(2)/2 only applies to a pure undistorted sine wave. Your scope is making an rms calculation for whatever waveform it has (distorted and asymmetrical as it may be) based on some set of sampled points, and in that process it must have some way of determining a zero reference. Given the analog nature of the times, I have to believe Mr D was speaking in reference to pure sine waves.

Good point! And suddenly, all those people who still swear by their analogue scopes, start to make sense.

The phase of the signal at the preamp out is determined by the number of inverting stages and any RC filters in the signal path. A basic low-pass filter shifts phase somewhere between 0 and -90 degrees depending on the frequency of the signal wrt the corner point of the filter. Likewise, a basic high-pass will shift the phase between +90 and 0 degrees. Of course there are several such filters along the path from the input to the preamp output, and if you change your test frequency you will see a different phase shift.

Excellent description - thank you! Phase shifting yes, but not necessarily distortion though? My maxed out signals are asymmetric saw-toothed. Is that the kind of thing you are used to seeing as a property of phase shifting?

[Stephen1966]

The Vrms/Vp relationship of Sqrt(2)/2 only applies to a pure undistorted sine wave. Your scope is making an rms calculation for whatever waveform it has (distorted and asymmetrical as it may be) based on some set of sampled points, and in that process it must have some way of determining a zero reference. Given the analog nature of the times, I have to believe Mr D was speaking in reference to pure sine waves.

Thinking about this some more, I can't say for sure what processing the scope applies when calculating Vrms and Vpp; if one is derived from the other, or if they are processed separately. I was looking for ways of filtering the noise in the signal and tried both changing the memory depth and reducing the sample rate but neither of these made any difference I could tell. I have an analogue scope as well which will give me a "digital" Vrms but there (with noisy signals) it's recommended measuring the Vpp by calculating the difference (referring to the graticules) from the top edge of the top peak to the top edge of the bottom peak. In analogue terms, this was as good as it was possible to get, with this earlier i.e. 80s scope. Vrms is the benchmark in any case so whether the maths agree on Vpp or not, it seems neither here nor there.

There are many environmental factors at play when taking measurements like these so it makes sense that he would round the measured figures off and regard it as a pure sine wave but given what I am seeing, I would go further, and say I believe he arrived at 2.5 Vrms after testing; not necessarily after working it all out on paper. Regarding the impedance of 500k though (after all, this was what this was all about in the beginning) it does seem likely that this refers to the expected impedance you might find in an effect but could it not also refer to the output impedance? In which case, it would probably be considered pretty poor but it is perhaps a coincidence that 500k is 50% of (an ideal) 1Meg Master pot. And in that position (which is pretty high on regular log taper pots) with an input signal around 100mV you get near enough 2.5Vrms coming off the Master's wiper.

Following up on sluckey's recommendations, I did some more testing with different input signal strengths using Merilin's charted figures for different kinds of pickups. 25mV for single coil/single notes; 50mV single coil strumming and humbucker single notes, and; 100mV for humbucker strumming.

Here is what 100mV input signal looks like with the volume and master, all the way up:

09 - 100mV - full.jpg

Here, is what it looks like with the volume and master at midnight on the dial (that's 30% with these j-taper pots):

08 - 100mV - half.jpg

And here, what it looks like putting out 2.5Vrms

07 - 100mV - 2.5Vrms.jpg

It is kind of curious that if I dialled up the volume and master pots to 1 or 2 o'clock on the dial, the pots would be around 50% of the 1Meg taper and it corresponds more or less, with 2.5Vrms at the Send jack. The sine is nice and smooth and even at full power, it isn't hugely distorted. I am thinking, as I mentioned before, the 2.5Vrms is a safe maximum operating limit for most low powered line out effects and from a practical perspective the volume and the master are quite high at the 2.5Vrms level. Stage level volume, easily. If you were using regular 10% taper pots, those positions of the master and volume would be even higher.

What I find interesting is how, if the amp is reined in properly, you get a smooth signal across a range of pickup types and levels, only the YouTube wankers seem interested in seeing what happens if they turn up their amps all the way and play! Testing circuits aside, we never play it like that in real life. I guess if you are using low output single coils from back in the day, you are going to get a higher SNR but you will get all the contours of the tone the pickups produce. And same, when you transition over to humbuckers. It's only the higher output modern pickups where you need to dial it back a bit. To get that signature tone from the amp it needs to be driven, but not too hard. I think this (as a for instance) is what he meant, when talking about setting the OD, you should approach it slowly. I believe this range of operating levels isn't a theoretical red herring, inaudible except to bats, but a real feature and sufficiently audible with a range of guitar and pickup types. Not talking so much about any preference for pickup types, rather the ability of the amp to let the guitar tone come through. The amp is incredibly touch and guitar tone sensitive in this sense. When you take into account that signature tone we get when we dial the amp in as well, we aren't dealing so much with a mere amp anymore; we can think of it as an instrument in its own right.

[pdf64]

When a 1M pot is set to 500k, the output impedance will be around 250k. Due to the output impedance of the stage feeding it being maybe 40k, rather than infinity.
Sorry if you mentioned it earlier, but where to buy j taper pots?

[Stephen1966]

When a 1M pot is set to 500k, the output impedance will be around 250k. Due to the output impedance of the stage feeding it being maybe 40k, rather than infinity.
Sorry if you mentioned it earlier, but where to buy j taper pots?

Aha! Because of the bonkers circuit I designed, I wasn't able to actually measure it. But good point and it takes us back to the idea that 500k is what the effect would present. Thanks Peter.

J-tapers came from TAD https://www.tubeampdoctor.com/en/a1m-lo ... eter?c=103. That was on the recommendation of another member here - can't remember who - but thanks

Edit: a PS to that, there is also the RC net we see in the Skyline #124 (220k with 250pF) which is active (in circuit) on the SEND jack when jacks are inserted. There's been some confusion about when this RC became active but this picture shows it isn't active when there is no return jack inserted.

ODS_124061.jpg

Further thoughts: the original 124 schematic (the 1984 version) didn't have this RC net on the SEND and RETURN jacks but the photo of the back of the chassis was screen printed with the same text. I don't think the chassis was changed or that the RC net was part of the original calculation. I think the net has a couple of effects: I guess it helps to attenuate the signal which is very hot (for all but line level effects) and it would also perform some of the functions of a low-pass shelving filter when you add in the capacitance of longer cables out. As I understand it, the parallel capacitor, theoretically limits the loss of higher frequencies in what would otherwise be a simple low-pass filter.