VOGONS


First post, by clb

User metadata
Rank Newbie
Rank
Newbie

Heya,

recently I learned something new about potentiometers that I did not know before, so thought to write a story about my experiences, and where my learning ended up with.

On a 8088 Turbo XT clone that I am restoring, there was an audio volume knob set up at the ISA slots on the back. Some 30 years out in cold storage had broken the potentiometer and it would no longer work properly. See the first picture "volume_knob_and_speaker.jpg".

The PC speaker looks like an 8Ohm impedance 2.25" speaker, has the writing "8Ohm 0.25W" on it. In particular, not the piezo buzzer. The speaker connects to pins 1 and 2 of the four-pin connector on the motherboard, and sounds great at max volume when there is no potentiometer connected in between.

Having a volume knob on the PC speaker seems like a great idea, and I thought that I'll do a great improvement and switch the volume knob to the front, so that adjusting audio volume will be really easy to do then. "I'll just go and order a suitable pot and all job done", I thought.

Ouch, turns out that sourcing a suitable pot seems like an impossible task.

First, measuring the old pot gave a resistance of ~10kOhms max, and given that this is an audio application, it seemed obvious that I'll order a 10kOhm logarithmic pot and that'll fit nicely.

When my order arrived and I tried it out, I would get good max volume at 100% of angle, and practically silent at 0% angle (though still tiny bit audible if one sits next to the machine and listens carefully, and there's no background noise), but surprisingly a sharp cutdown of volume when turning ten degrees or so, and already at 90% max setting, audio would be really quiet.

I was not really able to understand this, I thought that logarithmic pots (or "audio pots" as they are sometimes called) are exactly for this kind of volume setting purpose, to avoid the sharp audio volume cutoff that linear pot would have.

At first I thought that maybe I got the wrong pot and it was not a logarithmic pot after all. So I measured its resistance response curve between 0% and 100%, and got the leftmost graph in "pot_response_curves.png".

Shrugging it off, thinking that maybe I got confused somehow on the theory - maybe something is different since this is a PC speaker we are talking about and not "real" audio (whatever that distinction would mean...), I ordered a linear 10kOhm pot, thinking that that's what I'll need then.

Well... that potentiometer arrived and tested it, quickly realizing that it is no better - in fact - it was even worse than the logarithmic potentiometer: audio volume would fall off really sharply in a few degrees. Measured the linear pot's response curve, which naturally was linear. See the middle graph.

What on earth's going on?! Only here, (after also testing 5kOhm linear and log potentiometers, with only marginally better success) I actually paused to think what kind of response curve I'd really want to have to get a linear sounding audible response.What I would have liked was a pot that at the loud volume end, the resistance should change modestly, and at the quiet volume end, the resistance should change sharply.

Referencing that to the response from the logarithmic pot, I realized that if I wired the pot from the opposite side from the three legs, that would give me perfectly linear audible response, but with the adverse effect that the handedness would now be wrong: max volume would occur at the far left end, and minimum volume would be at the far right end, opposite to how volume knobs are generally expected to work. The response curve from this wiring is shown on the far right in the graph.

What I'd want is that same response curve, but mirrored to opposite handedness. Googling around, I learned that this kind of opposite handedness potentiometer is called "reverse logarithmic" or "reverse audio" or "anti-log" potentiometer - so that's what I'd need.

Got a 10kOhm reverse audio potentiometer, wired it in, and success! Well, 90% there. Handedness was now right, but I was a little bit underwhelmed about the volume range.. the PC speaker becomes quiet at maybe around 7kOhm or so. So the 10kOhm pot turned out to be a little bit too much, so that maybe 25-35% of the range at the end wasn't that useful. It works great, but since retro restoration projects like this end up seeking for perfection, I did not stop there.

That converged me to use a 5kOhm reverse audio potentiometer, which is just perfect! Well, 95% I'd say. The volume range is really nice, no sharp cutoff, perceptually linear, and the whole range gives a meaningful control. However at 0% volume, the 5kOhm resistance is not quite enough to completely mute the speaker. Possibly when the speaker is enclosed inside the chassis, it might sound practically mute (haven't tested closed chassis yet).

While I had been searching for different potentiometers, I learned about some pots that have on-off switches in them, such as this Taiwan Alpha 5kOhm Audio potentiometer seen in image "5k_switch_pot.jpg". The on-off switch resides at the left end, so that one can wire it that turning the whole pot off cuts off the circuit.

This kind of potentiometer with an on-off switch would be perfect! At 5kOhm the volume is really quiet, but still slightly audible, so switching the circuit off completely would give a nice way for complete audio control.

However, despite a lot of searching, I have not been able to find such a "5kOhm Reverse Audio Potentiometer With an On-Off Switch at the End" being manufactured anywhere 🙁 It even reads as theoretical like the elusive Rubber Chicken with a Pulley in the Middle. God damn.

So for now, I have had to settle with a 5kOhm Reverse Audio Potentiometer Without an On-Off Switch at the End. You can see the final image at "final_result.jpg". Drilled the pot into the CF<->IDE adapter chassis since I did not dare to drill into the original 5.25"->3.5" enclosure chassis. I'll add a "VOL" sticker below the knob, and it oughta stand out fine from the CF card.

I won't be able to completely mute the speaker on this build then, but why would I want to do such a blasphemy anyway 😜

I am still not 100% sure why this application needs a reverse logarithmic potentiometer rather than a general logarithmic potentiometer, but it works, so I'm happy.

Have you had a pot adventure similar to this? (phrasing..) Which item did you end up with?

Attachments

Reply 1 of 9, by BitWrangler

User metadata
Rank l33t
Rank
l33t

10K sounded too high to me for audio volume, most stuff I've seen it would be 1-5K... I wonder if it's not original even, so the part you found in there never "worked perfect" but was a "I'll just stick this in because it's what I've got kicking around".

Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.

Reply 3 of 9, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie

Made a log/lin mistake fixing a guitar once.

I'm curious how you're connecting yours up here, have you got any photos/diagrams showing what's connected where? The XT speaker header (from a schematic) was 1-Signal, 2-NC, 3-GND, 4-+5V, with pin 1 normally held at +5, then toggling from +5 to GND to make a beep. I've always used pins 1&4 for connecting a speaker. The signal apparently has a 33ohm resistor on it, so even if the output is shorted the current is limited to ~150mA, and the driver can handle 300mA.

I'm not sure if what follows is correct, but I think, assuming the resistor is in series with the speaker (so just using 2 pins of the resistor), then when it's at 0 ohms then there'll be very roughly 100mA (5V / ~~50ohm) through the speaker. When it's at 500 ohms then very roughly 10mA and at 5k, 1mA. I think that means (ballpark figures here) that the speaker power will be reduced by a factor of 100 for each factor of 10 the resistance increases. 100x the power is 4x the apparent loudness. So a 5k pot will give (very very roughly) a 16x range from quiet to loud, so about 40dB. There's probably some way to get it to turn fully off. Maybe connect the pot pins 1&3 to the signal and +5V, then pin 2 to one side of the speaker, and the other side of the speaker to +5V?

[edit: don't think that works. Still think it can be done]

Reply 4 of 9, by creepingnet

User metadata
Rank Oldbie
Rank
Oldbie

If the volume is not going down fully you would need to ground it off - this is coming from my Guitar Logic. Typically, looking at the bottom of the pot, from left to right, the hot lead (From PIT in this case) would go to this lug, middle lug would be output, and the right one would be grounded off to the POT casing and the pot grounded to the other lead the speaker goes through that would act as a ground. That's basically all a guitar does, is sends the hot signal to ground to turn the volume down.

However, I tried what you did with my Tandy 1000 by building an interposter to the speaker and that did not fully cut the volume down, I was using a 500K logarithmic you can buy at a guitar store.

~The Creeping Network~
My Youtube Channel - https://www.youtube.com/creepingnet
Creepingnet's World - https://creepingnet.neocities.org/

Reply 5 of 9, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie

Ok, that helps. So how much does anyone trust that the 33ohm current limiting resistor shown in the 5160 schematic actually exists? If it exists then it shouldn't matter if pin 1 (signal) is shorted to pin 4 (+5V). In which case something like this might work (there's a slider over on the right to adjust the pot):
https://www.falstad.com/circuit/circuitjs.htm … iS4i07UAe3cLCAA

Connect header pin 4 (+5V) to pot left and to speaker +. Connect speaker- to pot right. Connect pot middle to header pin 1. That way, pot turned fully left shorts the +5 to header pin 1, so no current flows through the speaker. Pot turned fully right then nearly all the current will flow through the speaker (there will be a small amount of current through the bulk of the pot). Shape of the curve will be set by both the curve of the pot, but also its value. When turning the volume down then a higher value pot will drop off quicker at the start, lower values will drop off quicker at the end.

Haven't tested it. If the 33ohm resistor isn't somewhere between pin 1 and the actual speaker output then this does end up connecting +5V directly to whatever chip is trying to drive pin 1 to Ground. Which would be bad for that component and any PCB traces. Could fit an external limiting resistor between header pin 1 and pot pin 2, but that'd reduce the max volume from the speaker.

Reply 6 of 9, by clb

User metadata
Rank Newbie
Rank
Newbie
BitWrangler wrote on 2022-01-28, 17:22:

10K sounded too high to me for audio volume, most stuff I've seen it would be 1-5K... I wonder if it's not original even, so the part you found in there never "worked perfect" but was a "I'll just stick this in because it's what I've got kicking around".

I think you are right there - looking at the old 10kOhm potentiometer on the back panel, I also got the vibe that it was not an original part of the PC build, but something that the original owner had themselves added in in the 80s.

Reply 7 of 9, by clb

User metadata
Rank Newbie
Rank
Newbie
snufkin wrote on 2022-01-28, 23:10:

I'm curious how you're connecting yours up here, have you got any photos/diagrams showing what's connected where?

Sure - I followed the original wiring, which connected two legs of the pot in series. Attached illustrations of the old and new wiring.

snufkin wrote on 2022-01-28, 23:10:

The XT speaker header (from a schematic) was 1-Signal, 2-NC, 3-GND, 4-+5V, with pin 1 normally held at +5, then toggling from +5 to GND to make a beep. I've always used pins 1&4 for connecting a speaker. The signal apparently has a 33ohm resistor on it, so even if the output is shorted the current is limited to ~150mA, and the driver can handle 300mA.

I find that the manual at http://minuszerodegrees.net/manuals/Turbo-XT% … ion%20Guide.pdf describes the board that I am working on here. On page 19, that manual lists the PC speaker connection as

Pin 1: Data out
Pin 2: 35 Vdc
Pin 3: Ground
Pin 4: +5 Vdc

(attached screenshot)

That 35 Vdc is news to me - I wonder if that might be an OCR scanning problem (there seem to be some of those in that document). The old PC speaker was connected to pins 1 and 2, so Pin 2 can't be unconnected at least.

When I take a multimeter to the pins when the computer is running, pins 3-4 do measure ~5 volts, but oddly pins 3-2 measure zero.

Huh, and also when I measured pins 1-3 (or 1-2, actually not sure which ones I was exactly on there), a power relay triggered off on the PSU and the computer and it shuts off. (I have a new ATX PSU with an ATX->AT adapter here) Nothing seems to be broken though, it happily boots back up and PC speaker is still working.

Attachments

Reply 8 of 9, by clb

User metadata
Rank Newbie
Rank
Newbie
snufkin wrote on 2022-01-29, 01:50:

Ok, that helps. So how much does anyone trust that the 33ohm current limiting resistor shown in the 5160 schematic actually exists? If it exists then it shouldn't matter if pin 1 (signal) is shorted to pin 4 (+5V).

Connect header pin 4 (+5V) to pot left and to speaker +. Connect speaker- to pot right. Connect pot middle to header pin 1. That way, pot turned fully left shorts the +5 to header pin 1, so no current flows through the speaker. Pot turned fully right then nearly all the current will flow through the speaker (there will be a small amount of current through the bulk of the pot). Shape of the curve will be set by both the curve of the pot, but also its value. When turning the volume down then a higher value pot will drop off quicker at the start, lower values will drop off quicker at the end.

Haven't tested it. If the 33ohm resistor isn't somewhere between pin 1 and the actual speaker output then this does end up connecting +5V directly to whatever chip is trying to drive pin 1 to Ground. Which would be bad for that component and any PCB traces. Could fit an external limiting resistor between header pin 1 and pot pin 2, but that'd reduce the max volume from the speaker.

Doing a three-legged connection instead of a naive two-legged connection in series might make sense. Maybe I'll play with that by adding a safety resistor in between so I don't short pins 1 and 2/4 directly.

Also I wonder if this speaker should work if I connect it to pins 1 and 4 instead of pins 1 and 2.

Reply 9 of 9, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie

Nice, got to like manuals with schematics. So it looks like the header is wired: 1-Signal (switches between NC and 0V), 2-+5V, 3-GND, 4-+5V. So that 35 was a typo for +5. So accidentally shorting 2-3 or 3-4 will trip the PSU. It may also trip if 1-2 is shorted and it tries to beep.

It looks odd that the output from U85 pin 8 goes directly to the base of the speaker output NPN transistor. I just checked a 74LS00 datasheet and it says 0.4mA, although it later says that for periods of <1sec is can source anything from 20mA to 100mA in to a short (which is roughly what the B-E junction is). I did wonder if that output transistor would act to limit the current through the speaker, but if U85 is pushing 20mA in to the base of that transistor then it won't.

Regardless, the circuit I drew won't work. Ho hum. I'll see if anything else comes to mind.

Attachments

  • XT_Turbo_SpeakerOut.png
    Filename
    XT_Turbo_SpeakerOut.png
    File size
    56.62 KiB
    Views
    185 views
    File license
    Fair use/fair dealing exception