Figuring out reason for various resistor values on Potentiometer circuits

I may have an answer to why 1.5K resistors are put across the pots . If you make the pots ( I use the word pots but they look more like straight possibly wire wound potentiometer tracks ) 1K ohm then with a 1.5K resistor in parallel you end up with a 600 ohm output impedance which is a standard transmission line impedance to connect to the filters . As for the use of different resistor values in the attenuators I do not know all I can say is there combined resistance offers an approximate 800 ohm load to the sources .


Can you explain why it says in the franklin institute article , the output from the Potentiometers is logarithmic ? If the 1.5K resistor isn't there to make it logarithmic , what is making this keyboard circuit logarithmic? Unless they are using Logarithmic Potentiometers?
 
I did check the potentiometers in an online spice program, and this is what I found out. If I use a 1K Linear POT with a 1.5K in parallel resistor like what epicyclic says the POT value most likely was the output is mostly linear like, If I use a different potentiometer value that is still Linear like 15K , 20K , 100K , etc , with the 1.5K parallel resistor , the output is logarithmic like.
 

PRR

Member
Joined 2003
Paid Member
Here is a graph I have drawn , it may or may not be useful .

Partly helpful. The sound source, a rude buzz, is not "flat" spectrum, nor do we really want "flat with F-M loudness correction". It may also be some inverse of the mouth's many resonances.

It is obviously not make-or-break if you want to DIY it. Make them all the same. A few will grate on the ear, turn those down (reduce their maximum output). Now a few more will grate, adjust. You may find the extreme bass won't come out, limiting faithfulness for bass-baritone voices; boost bigtime or let the bass be weak, 100 years of telephones sloped-off below 300Hz.
 
Regarding using a reverse 1K pot with the 1.5K load resistor I agree it does not give a good control range , I got seduced by the resulting 600 ohm paralleling result and something like 15K works better with the 1.5K . Now here is an additional hurdle if you want to get really close to the original design , what is the input impedance of the filters ! as anything below about 15K ( input Z ) will start to modify the 1.5K loading significantly .


Onto the logarithmic problem.......... The Franklin article says the voltage increases approximately logarithmically which is different from saying we used a logarithmic pot . A linear pot used the normal way around and used as a volume control gives an initial burst of volume when turned from the minimum position because the dB spread is bunched up at the beginning . A linear pot used in reverse with a finite load as is being used in the Voder reverses this bunching so that the dB range bunching occurs at the max volume position and you end up with an approximate logarithmic voltage progression on the output . This is where the crossed lines has occurred as an ideal logarithmic pot gives a linear dB progression IE say a 20dB spread then the increase in dBs is spread evenly over the range of the pot, like 0 , 5, 10, 15 , 20 .
 
Last edited:
Regarding using a reverse 1K pot with the 1.5K load resistor I agree it does not give a good control range , I got seduced by the resulting 600 ohm paralleling result and something like 15K works better with the 1.5K . Now here is an additional hurdle if you want to get really close to the original design , what is the input impedance of the filters ! as anything below about 15K ( input Z ) will start to modify the 1.5K loading significantly .


Onto the logarithmic problem.......... The Franklin article says the voltage increases approximately logarithmically which is different from saying we used a logarithmic pot . A linear pot used the normal way around and used as a volume control gives an initial burst of volume when turned from the minimum position because the dB spread is bunched up at the beginning . A linear pot used in reverse with a finite load as is being used in the Voder reverses this bunching so that the dB range bunching occurs at the max volume position and you end up with an approximate logarithmic voltage progression on the output . This is where the crossed lines has occurred as an ideal logarithmic pot gives a linear dB progression IE say a 20dB spread then the increase in dBs is spread evenly over the range of the pot, like 0 , 5, 10, 15 , 20 .


I have no idea what the input impedance of the filters are , I would need to ask the person who has a copy of the blueprints for the voder.

I do know some specs of the filters, Like for example each filter are 2 2-pole filter sections that are slightly overlapping with each section having a minimum Q of 10. But another question is though, since most filters are now made with op amps , what is the usual input impedance of an op amp?

Instead of using the key circuit from the franklin article/ circuit from the original blueprints. Could I just do the potentiometers the normal way? Using a standard linear pot, and the wiper of the pot as the output , with a resistor from the wiper to ground to make it Logarithmic like? It does seem in the patent for the voder , this is the way they were going to it. But I don't understand why they had to do it the way they did it.

The person in the youtube video who made the clone told me he is going to add the resistor voltage divider circuit in the beginning of the key circuits using potentiometers , so he Can adjust the Audio output of each key. This is probably the way I am going to do it as well.
 
I would stick with the Franklin configuration as it says they spent a lot of time fine tuning the Voder by ear ( this is the way they achieved the best results rather than trying to copy the oscillograms ) . If you use a log pot or a log converted linear pot in the normal manner you will not be copying the characteristics of the Voder as explained in the Franklin article .The choice is obviously yours , you could try both methods . Using a log type pot in the normal way may or may not give better results . When I do not know and cannot find out the why I copy the documented working circuit , which in this case appears to be associated with the Franklin article .

Filter loading of the pots need not be a problem as now you would use a buffer between the filters and the pots which ever version of pot you used . I only mentioned it because the YouTube video says they used every day telecoms equipment and this says to me the possibility of standard 600 ohm impedances being involved which would modify the choice of pot value .
 
Last edited:
Do you have construction details of the actual Voder key assemblies , I am not asking you to share them . Its just that I foresee problems with getting full rotation of an ordinary potentiometer . I am pretty sure the clone on YouTube does not achieve this as he has mounted the pots at the pivot point .
 
Do you have construction details of the actual Voder key assemblies , I am not asking you to share them . Its just that I foresee problems with getting full rotation of an ordinary potentiometer . I am pretty sure the clone on YouTube does not achieve this as he has mounted the pots at the pivot point .

I don't have any construction details on the Voder Key assemblies.

You are right about the clone on youtube with its keys not getting a full rotation , when you press down a key on that clone, it only goes to 50% of the potentiometer.

I do have various fixes for this problem though I think. one way was to use a 1:2 gear ratio system , have a potentiometer with cog wheel on it, so if you press down completely on the key it will turn the pot 2x as fast , making it go 100%.

Another option I think would work is to use a Linear Displacement Sensor :

https://www.mouser.com/ProductDetai...tronics/404R5KL10?qs=LUfMuE0iDoqgEWMHtVSIag==
 
Yes the linear motion potentiometer could work quite well provided the force to move it is not high ( I am thinking about the feel of the key for the operator ) . You could mount the pot directly under the finger area part of the key to achieve full movement . I think you would need a bearing cup under the key or some kind of linkage to cover the transition of the arch movement to linear motion .
 
Actually no , invert the linear motion potentiometer and mount it to a Voder type key frame at the other end of the seesawing key assembly . This would make mounting easier and give far more access for adjustment .

Actually that is the correct way to use those potentiometers , The person who has a copy of all the original blueprints , also made the 1st clone of the voder back in late 2015, and that is what he actually used. It's actually 2 clones, one clone was built for bell labs, and the 2nd one is his portable personal one.

0428171712a-HDR.jpg