I'll keep this simple.
This is a side line topic from my gu81 se amp build.
After a lot of reading on output transformers and also looking up parafeed combinations I'm left rather confused. But this is what I can deduce...
-Plate load choke needs to be like a normal opt transformer so say approx 50-100h. What determines this value?
- the opt transformer remains the same impedance as non parafeed combinations. I'm no engineer and I'm assuming the plate sees the capacitor as a short circuit at audio frequencies, hence the opt remains the same
-coupling capacitor should fall between 3uf-8uf
Now I'm not looking for perfection yet.... Just a working combination/prototype
Tim
This is a side line topic from my gu81 se amp build.
After a lot of reading on output transformers and also looking up parafeed combinations I'm left rather confused. But this is what I can deduce...
-Plate load choke needs to be like a normal opt transformer so say approx 50-100h. What determines this value?
- the opt transformer remains the same impedance as non parafeed combinations. I'm no engineer and I'm assuming the plate sees the capacitor as a short circuit at audio frequencies, hence the opt remains the same
-coupling capacitor should fall between 3uf-8uf
Now I'm not looking for perfection yet.... Just a working combination/prototype
Tim
Parafeed is also called choke feed - so best to google search on both terms. I know Bottlehead and Patrick Turner's website have designs (Patrick has passed, so his original website has gone but there is a mirror somewhere). You need to be aware of the various LC networks and their impact on design.
Choke feed, ok I'll search that as I've always gone down the parafeed search path. Cheers,
yes Patrick Turner was a amazing source of info. Being one of the "young guys' (nearly 40) I'm often finding out these people have passed away after I discover their webpages/legacy.
Knowing a bit about x over design, helps me with the LC component.
Where I'm at for now is making something work and then gaining an understanding of how/why. There's so many people with plenty of advice based on theory, I'm hunting for real world examples. Patrick Turner was one of the makers and thinkers.
Thanks again
Patrick was able to prepare some choke-feed amps as he could wind the choke to suit the application. Depending on what power and frequency response you are after, the choke can be an expensive custom build if you can't make it yourself.
I prepared and tested a special test amp that needed a very low frequency response, and with feedback, and could only deploy about half the choke inductance needed - that meant I couldn't roll off the LF end suitable for stability, so am biding my time to come across more suitable chokes (with high L/R ratio).
The coupling cap can also be onerous if your application can't cope with e-cap leakage current - luckily I had large poly's, as the first effort with ecaps was an issue.
I prepared and tested a special test amp that needed a very low frequency response, and with feedback, and could only deploy about half the choke inductance needed - that meant I couldn't roll off the LF end suitable for stability, so am biding my time to come across more suitable chokes (with high L/R ratio).
The coupling cap can also be onerous if your application can't cope with e-cap leakage current - luckily I had large poly's, as the first effort with ecaps was an issue.
The good news is I have plenty of large Polly caps, 10-40uf 1000v (salvaged from huge grid tie inverter). Choke winding Is limited by my patience, of course I don't have the ability to plough into opt winding. Patrick's info is close on hand as I will attempt to wind some opts in the near future but maybie not on this job
Experiences like you've mentioned about choke inductance bring too low limits LF cutoff frequency and leads to instability is great advice.
Experiences like you've mentioned about choke inductance bring too low limits LF cutoff frequency and leads to instability is great advice.
The best value for the plate choke may be "infinity". But infinite chokes are infinitely out of stock.
The next path to try is Filter Theory. You have a source resistance, a shunt choke, a series capacitor, a loaded choke (transformer and speaker). If you pick these values poorly you get a saggy or peaky response. Do it well and you have a 3rd-order high-pass with Q to your taste.
The Dead Men figured this all on slide-rule. From books.
The next path to try is Filter Theory. You have a source resistance, a shunt choke, a series capacitor, a loaded choke (transformer and speaker). If you pick these values poorly you get a saggy or peaky response. Do it well and you have a 3rd-order high-pass with Q to your taste.
The Dead Men figured this all on slide-rule. From books.
Last edited:
I have done similar simulations to those in PRR's post. As a practical matter, I size the parafeed cap to be
C = 2*L/(R*R)
L is plate choke inductance, R is output transformer primary impedance. This keeps the load impedance seen by the tube both resistive and fairly constant down to the lowest practical frequency. Around that point, it's not too sensitive - anything from half to twice the calculated value is reasonable.
Assuming the amp is a SET, larger capacitance will extend the bass extension but present a low impedance load to the tube, limiting available power.
C = 2*L/(R*R)
L is plate choke inductance, R is output transformer primary impedance. This keeps the load impedance seen by the tube both resistive and fairly constant down to the lowest practical frequency. Around that point, it's not too sensitive - anything from half to twice the calculated value is reasonable.
Assuming the amp is a SET, larger capacitance will extend the bass extension but present a low impedance load to the tube, limiting available power.
I am back onto this job now (Easter break is on so I'm away from shed but have plenty of time to read/research) I'm intending do this:
I have a pair of se opts 2.5k-8ohm 10w... 100ma? (That's what they told me) In a past life I loaded them up on a 811a A2 amp 8 ohms on the 4 ohm winding probably 5k as seen on the plates of the 811s. they had good bass and sounded really sweet.
If I have a go with these in parafeed I'd expect them to handle a lot more power say 30-50w. But would actually this apply? and how about the end gap in the lams..
From a copper perspective they have heavy enough wire for the power, from a lamination stack view point it's easily equivalent to a 50-100w pp transformer. Any ideas about this?
I have a pair of se opts 2.5k-8ohm 10w... 100ma? (That's what they told me) In a past life I loaded them up on a 811a A2 amp 8 ohms on the 4 ohm winding probably 5k as seen on the plates of the 811s. they had good bass and sounded really sweet.
If I have a go with these in parafeed I'd expect them to handle a lot more power say 30-50w. But would actually this apply? and how about the end gap in the lams..
From a copper perspective they have heavy enough wire for the power, from a lamination stack view point it's easily equivalent to a 50-100w pp transformer. Any ideas about this?
PRR,
Thanks for reminding me of what I call the "Filter Bible".
A definitive work all about filters.
Mr. Zverev, a Russian, who worked for Westinghouse wrote that book.
Over 30 years ago, that book sold used for $208.
The chief engineer where I worked was my mentor. He had me use that book, the methods and tables in the book, to design a 525 MHz 5 pole Helical filter for a spectrum analyzer IF channel.
The model shop showed me and authorized me to use the milling machine (they got tired of me wanting the coupling walls cut down multiple times at 50 thousandths at a time), I made the required measurements after each cut. There were lots of things to learn, including calculating the taps on the input and output helical resonators.
I escaped the hard work of using a slide rule, instead I used an in-house software, entering Zverev's table values.
With the 8088 processor, I could run the program, go get coffee at the cafeteria, stop by one of the engineers desk for a short talk, and when I got back, the calculations were just finishing up.
Then, I got my boss to have an 8087 co-processor put in, and now the 8087 version of the software made short order of those calculations.
Slide Rules . . . for the most part, I did not use them since college chemistry and physics; but my own electronics designs were the exceptions until I sprung for an HP-11 calculator in 1985, it still runs today.
One thing about the ideal filter design, is that the final load at the end of all of the output circuit of a Parafeed amplifier is . . .
A loudspeaker, and so many of them have impedances and phase angles that vary all over the map, versus frequency.
In some ways, controlled RF impedances, and waveguides, etc. were easier to deal with.
But that was just a reflection on me (both pun and reality intended).
Thanks for reminding me of what I call the "Filter Bible".
A definitive work all about filters.
Mr. Zverev, a Russian, who worked for Westinghouse wrote that book.
Over 30 years ago, that book sold used for $208.
The chief engineer where I worked was my mentor. He had me use that book, the methods and tables in the book, to design a 525 MHz 5 pole Helical filter for a spectrum analyzer IF channel.
The model shop showed me and authorized me to use the milling machine (they got tired of me wanting the coupling walls cut down multiple times at 50 thousandths at a time), I made the required measurements after each cut. There were lots of things to learn, including calculating the taps on the input and output helical resonators.
I escaped the hard work of using a slide rule, instead I used an in-house software, entering Zverev's table values.
With the 8088 processor, I could run the program, go get coffee at the cafeteria, stop by one of the engineers desk for a short talk, and when I got back, the calculations were just finishing up.
Then, I got my boss to have an 8087 co-processor put in, and now the 8087 version of the software made short order of those calculations.
Slide Rules . . . for the most part, I did not use them since college chemistry and physics; but my own electronics designs were the exceptions until I sprung for an HP-11 calculator in 1985, it still runs today.
One thing about the ideal filter design, is that the final load at the end of all of the output circuit of a Parafeed amplifier is . . .
A loudspeaker, and so many of them have impedances and phase angles that vary all over the map, versus frequency.
In some ways, controlled RF impedances, and waveguides, etc. were easier to deal with.
But that was just a reflection on me (both pun and reality intended).
Last edited: