Welcome! Need support, you got it. Or share your ideas and experiences.
Return to MagneQuest/Peerless Forum
70.117.113.124
This forum seems to have a lot of folks that are into parafeed, so thought I'd post this here. I'm wondering about folks experience and opinions on the different ways to parafeed.AFAIK there are 4 different ways to parafeed with a single ended amp, #1, I'll call conventional, which uses the parafeed cap between the tube anode, and the primary of the OPT. #2, There is a variation of this which moves the parafeed cap to the ground leg of the OPT, then to ground. #3, I'll call Western Electric Output, because that is what I've seen Lynn Olson call it. With that topology one lead of the parafeed cap is on the ground leg of the OPT, and the other lead goes to the top of the power tube cathode, with an un-bypassed cathode resistor going to ground. #4, The same topology as #3 except the cathode resistor is bypassed.
I've built 2 parafeed projects, one a line stage, and another an integrated bread boarded 6p36s SET. With the line stage (6n6p, CCS loaded, Magnequest B-7 15k nickle OPT) I pretty much evaluated everything by my own preferences. I know this is pretty subjective, so take it for what it may or may not be worth. I tried all of the parafeed variations, with four different biasing schemes. The biasing schemes were cathode resistor bias with cathode bypass cap, un-bypassed cathode resistor, cathode battery bias, and cathode diode bias. The topology I ended up preferring was Western Electric Output, with diode bias.
The 6p36s integrated SET used a 6e6p-dr driver stage (diode bias, CCS loaded), cap coupled to a parafeed 6p36s power stage (choke loaded, Magnequest nickle core TFA-2004JR OPT). Experimentation was carried out both subjectively, and with an oscilloscope and signal generator. Subjectively I decided on the 2 versions of Western Electric Output sounded best, and wanted to measure the power and bandwidth of that topology with a bypassed and un-bypassed cathode resistor with the scope.
First we measured the version with the cathode bypass capacitor. We were surprised to find that the power was 8 watts (theoretically, should have been 6 watts at the most), the bandwidth was flat from 23K to 10hz. Next we tried it without the cathode bypass cap. Output was only down by 1 watt, to 7 watts. The bandwidth on the high end was the same 23k, but was flat down to only 30hz, about 20hz higher than the previous topology.
My speakers are pretty flat down to 30hz, and when I tried switching the cathode bypass cap in and out of the circuit, I couldn't hear any difference. Since I didn't seem to need the cathode bypass caps, I decided not to use them.
I know my experience differs from others. I just read in a previous post that Mike prefers the conventional way to parafeed. I've also read from a couple of very knowledgeable folks, who recommend that with the Western Electric topology, the use of a cathode bypass cap is necessary for bias stability.
Anybody care to share their thoughts and experience on parafeed topologies? I'm wondering why the Western Electric feed is not more popular. Sure seems to be a win situation if we can eliminate that big honking cathode bypass cap. Any thoughts about cathode bias stability? I'm not clear on how that works.
twystd
Edits: 05/13/15 05/13/15Follow Ups:
I have tried conventional choke loaded parafeed SE 2A3 and active CCS loaded parafeed with SE 45, triode 6V6 & 6CB5.
If you can dig up the old VoltSecond paper on extended C4S, try that with SE 45 finals using a high mu/jump factor driver stage.
The -3dB went out to 37KHz with inexpensive Altec 70V line OPT with the 45 finals.
I find that active parafeed knocks down the ripple a lot and allows for smaller uF values in the power supply filtering, but you will still have a quiet amp circuit.
Can't seem to find the VoltSecond paper, sure would like to read it. When I google it, I find references to it being on the loose parts section of the Bottlehead site, but Doc has changed his site, so I can't find that section anymore. BTW, When I say the high end response was flat to 23K I have no idea where the -3dB point was, that was just where we could detect any drop off of response.By active parafeed, I assume you are talking about parafeed using a CCS, rather than a choke. The reason I didn't try that with my 6p36s project was that I was running it at 20 watts dissipation. I know that is considered too high by many, but I lurked on some Russian forums, and they commonly run them between 18 and 22 watts in triode mode. At my operating points, the tube is drawing 70ma. With the voltage swing of the tube (actual mu of~6.5), plus another 50V for compliance, heat sinking the CCS becomes a real problem. I figure I'd have to heat sink for about 30 watts for a safe margin. I did do a CCS parafeed on my 6n6p line stage.
The way I figure the minimum voltage for my power supply when using CCSs as parafeed loads, is to add the idle voltage, plus the mu x the bias voltage, then add another 50V for compliance. Any extra is just gravy as long as you can heat sink for the dissipation.
BTW, good point about less filtration necessary, that's always a good thing.
twystd
Edits: 06/10/15
I have also measured the reduction in bass extension with an unbypassed cathode resistor. To over-simplify, if you assume the parafeed cap is small than at low frequencies the rolloff point is set by the tube's rp and the plate choke inductance. With a unbypassed cathode resistance, the rp is increased significantly, increasing the LF corner.
Obviously it gets more complicated when you add in the real parafeed capacitance. I have also made simulations which show the importance of including the plate choke and output transformer core losses if you are not to draw unwarranted conclusions.
In spite of that, I will note that the unbypassed cathode resistor, by increasing the effective rp, reduces the damping factor in the deep bass, strongly affecting the bass of most speakers. The interactions here are too complex to simplify into rules of thumb, so experimentation and careful listening is still the order of the day!
Thanks Paul, lucid comment as usual. When I did the subjective evaluation, I used a pair of speakers that were very good in the middle and highs, but basically had no bass. These amps are designed to go full range. I'd be very disappointing if they had flabby bass when I hooked them up to full range speakers. It's one thing to measure a frequency with a resistive load and an Oscope, and quite another to hear if that bass is tight and fast with real speakers.I guess that could explain why that topology is not more popular. I will have to revisit the situation with full range speakers. Thanks again for your comment.
twystd
Edits: 05/15/15
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: