Gyraf / Purusha pultec frequency shift across all bands

[atticmike]

Hey fellas,

Anybody familiar with the gyraf pultec Purusha design and might be able to tell me how I would be able to compensate for a 1k shift across all bands?

Mike

PS: Increase the capacity on a band I don't need?

 

[gyraf]

Very unclear what you're asking..

And I never did anything with Purusha iirc??

Jakob E.

 

[shabtek]

likely referring to the HZ labels on purushas chassis not lining up with measured fc.

check out Ians poorman pultec documentation, he explains the circuit completely.

 

[okgb]

Should not associate his name with Gyraf's , he sells cases , ask him to help you.

 

[atticmike]

yeah sorry gyraf, didn't mean to affiliate you with his filter board.

does anybody know the value i'd need for 1k and 1.5k capacitance wise?

 

[atticmike]

well, I replaced the 16k band I don't need anyways with 150n and that's what happened

one band lower at 14k looks like this

i thought it might be logarithmic so since 78n is supposedly around 1k (which it is not, it is 1.9k for me), I added 150n

 

[atticmike]

replaced the 14k cap with 100n and now I'm getting a similar broad at 3.8k

what am I missing? this band connected with the wrong inductance?

 

[ruffrecords]

replaced the 14k cap with 100n and now I'm getting a similar broad at 3.8k

what am I missing? this band connected with the wrong inductance?

All these bands are VERY broad. The Pultec hi boost is meant to be much sharper. Check the value of the bandwidth pot you have fitted.

Cheers

Ian

 

[atticmike]

replaced the 14k cap with 100n and now I'm getting a similar broad at 3.8k

what am I missing? this band connected with the wrong inductance?

All these bands are VERY broad. The Pultec hi boost is meant to be much sharper. Check the value of the bandwidth pot you have fitted.

Cheers

Ian

yeah i think I already know why it will not work because I used the inductance from 16k with a value from 312H.

Could you help me do the math for the following things:

with 78n at 312mH i'm getting 1.73 khz = what capacitance would be required to acquire 1k? I figure it'd be something like 200n, roughly by my childish mental math

is it possible to approximate without me abusing the board by soldering on and off to figure the values out?

Already got pretty close by running 100n / 150n parallel with the existing 78n, being at 1.3k.

78n at 312mH is as above mentioned 1.73 khz = goal is 1k (I got to 1.3k with 78n + 100n /150n, don't remember whether it was 100n or 150n, either of them)
37n at 312mH is 2.5kHz = goal is 1.5k
39n at 155mH is 3.15kHz = goal is 2k
22n at 155mH is 4.15kHz = goal is 3k
20n at 78mH is 4.7kHz = goal is 4k

I'd love to figure it out by soldering stuff on and off but I already abused the solder lugs / holes a lot

in case you got wolfram, you could share the math with me where I'd only be able to change the value and see how it's done

 

[ruffrecords]

Could you help me do the math for the following things:

with 78n at 312mH i'm getting 1.73 khz = what capacitance would be required to acquire 1k? I figure it'd be something like 200n, roughly by my childish mental math

is it possible to approximate without me abusing the board by soldering on and off to figure the values out?

Already got pretty close by running 100n / 150n parallel with the existing 78n, being at 1.3k.

78n at 312mH is as above mentioned 1.73 khz = goal is 1k (I got to 1.3k with 78n + 100n /150n, don't remember whether it was 100n or 150n, either of them)
37n at 312mH is 2.5kHz = goal is 1.5k
39n at 155mH is 3.15kHz = goal is 2k
22n at 155mH is 4.15kHz = goal is 3k
20n at 78mH is 4.7kHz = goal is 4k

I'd love to figure it out by soldering stuff on and off but I already abused the solder lugs / holes a lot

in case you got wolfram, you could share the math with me where I'd only be able to change the value and see how it's done

Check this out - firmula and on line calculator in one:

http://www.1728.org/resfreq.htm

For example, inputting 22nF and 155mH gives just over 2.7KHz

If you don't get these frequencies then I would suspect the inductor.

Cheers

Ian

 

[atticmike]

Could you help me do the math for the following things:

with 78n at 312mH i'm getting 1.73 khz = what capacitance would be required to acquire 1k? I figure it'd be something like 200n, roughly by my childish mental math

is it possible to approximate without me abusing the board by soldering on and off to figure the values out?

Already got pretty close by running 100n / 150n parallel with the existing 78n, being at 1.3k.

78n at 312mH is as above mentioned 1.73 khz = goal is 1k (I got to 1.3k with 78n + 100n /150n, don't remember whether it was 100n or 150n, either of them)
37n at 312mH is 2.5kHz = goal is 1.5k
39n at 155mH is 3.15kHz = goal is 2k
22n at 155mH is 4.15kHz = goal is 3k
20n at 78mH is 4.7kHz = goal is 4k

I'd love to figure it out by soldering stuff on and off but I already abused the solder lugs / holes a lot

in case you got wolfram, you could share the math with me where I'd only be able to change the value and see how it's done

Check this out - firmula and on line calculator in one:

http://www.1728.org/resfreq.htm

For example, inputting 22nF and 155mH gives just over 2.7KHz

If you don't get these frequencies then I would suspect the inductor.

Cheers

Ian

Could we create an offset math because I can't do anything about the faulty inductor :/ (tolerance)

 

[ruffrecords]

Could we create an offset math because I can't do anything about the faulty inductor :/ (tolerance)

The simplest thing to do is used the measured frequencies to work out what the actual value of the inductor is - you can use that same web page to do it. Then use the new inductor values to calculate your new capacitor values.

However, before you do that I would say it is quite uncommon for inductors to be that far out. The most common reason I have seen in the past for wrong frequencies and altered bell shapes is putting the inductor in the wrong way round. Are you absolutely sure you didn't do that?

Cheers

Ian

 

[atticmike]

Could we create an offset math because I can't do anything about the faulty inductor :/ (tolerance)

The simplest thing to do is used the measured frequencies to work out what the actual value of the inductor is - you can use that same web page to do it. Then use the new inductor values to calculate your new capacitor values.

However, before you do that I would say it is quite uncommon for inductors to be that far out. The most common reason I have seen in the past for wrong frequencies and altered bell shapes is putting the inductor in the wrong way round. Are you absolutely sure you didn't do that?

Cheers

Ian

I'd be surprised if I got such a small offset then?

going by the site, I got around  110 MicroHenrys, which is a third of what's needed :/

 

[ruffrecords]

Show me how you got the 110 microHenries.

Cheers

Ian

 

[atticmike]

Show me how you got the 110 microHenries.

Cheers

Ian

110  MilliHenrys

 

[atticmike]

got the bands working but now my Qs are wider, useless

I wonder how I possibly managed to chop the mH into half

the inductor sits right btw

BEFORE:

CORRECTD:

those are all the narrowest Q

 

[ruffrecords]

The Q of those curves look to have a Q of around 1 which is the sort of value you would expect from a Pultec set to the wide bandwidth setting. How do they vary with the bandwidth control setting?

Cheers

Ian

 

[atticmike]

The Q of those curves look to have a Q of around 1 which is the sort of value you would expect from a Pultec set to the wide bandwidth setting. How do they vary with the bandwidth control setting?

Cheers

Ian

If I increase them, they become wider.

which one you say is the widest for a pultec, bottom picture? or the top one already?

 

[ruffrecords]

The Q of those curves look to have a Q of around 1 which is the sort of value you would expect from a Pultec set to the wide bandwidth setting. How do they vary with the bandwidth control setting?

Cheers

Ian

If I increase them, they become wider.

which one you say is the widest for a pultec, bottom picture? or the top one already?

I used the bottom ones to estimate the Q. It is quite siple to do an you can probably do it better than I can if you have a cursor that can read out the actual frequencies. All you need to do is measure the frequency above the peak where the response frops by 3db, then measure the point below the peak where the response drops by 3dB. Take the difference of these two frequencies and divide by the peak frequency. The answer is the Q

If The peaks get wider than this then you need to check the values of the bandwidth pot and the resistor in series with it. One of these is probably wrong.

Cheers

ian

 

[atticmike]

The Q of those curves look to have a Q of around 1 which is the sort of value you would expect from a Pultec set to the wide bandwidth setting. How do they vary with the bandwidth control setting?

Cheers

Ian

If I increase them, they become wider.

which one you say is the widest for a pultec, bottom picture? or the top one already?

I used the bottom ones to estimate the Q. It is quite siple to do an you can probably do it better than I can if you have a cursor that can read out the actual frequencies. All you need to do is measure the frequency above the peak where the response frops by 3db, then measure the point below the peak where the response drops by 3dB. Take the difference of these two frequencies and divide by the peak frequency. The answer is the Q

If The peaks get wider than this then you need to check the values of the bandwidth pot and the resistor in series with it. One of these is probably wrong.

Cheers

ian

the thing i don't understand is why i get third of mH for 312 and less in general from all

could the Q pot shift the mH value? I'm using a 5k because I was told it'd give me a wider range.