nickel in PS chokes, plate chokes, and output trannies?

Design and use of Chokes for PS, anode and filament use.
Post Reply
artsybrute
Posts: 6
Joined: Mon Jan 31, 2005 2:57 am

nickel in PS chokes, plate chokes, and output trannies?

Post by artsybrute »

Hi Dave and all,

Finally got around to it: Here are the present and proposed preamp schematics that pertain to magnetic upgrades.

All in all, the pre is working but with a few problems. Firstly, upon replacing out the last 380 uF electrolytic cap in the HT PS with a 100uF oiler, ripple rose from 0 to 12mV and it is clearly audible, as it gets in the way of the phono stage. That should get cleared up by the HT PS change shown, ie. putting a new nickel(?) choke between the 100 mike and 30 mike caps. While we're at it, might as well change the 80 rectifier to 5ar4, bringing up the voltage, which slightly reduces the value needed for the cathode bypass caps by increasing RL.

Those caps need to be increased about ten-fold, and I think that is the reason for some midrange raspiness at this point. I hope so.

Also, since the first two plate chokes are only rated for 10ma and are flowing 13.2, I worry about saturation. How would nickel lams change the sound here (assuming new chokes rated at 15ma or so)?

Third, is there a way to try nickel lams in place of the cobalt ones for the output trannies? As shown on the prints, frame height is 1-5/16" with 2 inches between bolt hole centers.

Last, there just seems to be way too many capacitors in this design. Any ideas on trimming some out without losing sound quality?


TIA,

Len
Attachments
now_118.pdf
higher res PDF of current
(322.26 KiB) Downloaded 539 times
proposed_193.pdf
higer res PDF of changes
(331.08 KiB) Downloaded 567 times
with proposed changes
with proposed changes
proposed_193.gif (55.38 KiB) Viewed 14338 times
As it is now
As it is now
now_118.gif (53.87 KiB) Viewed 14339 times
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

hey len, i shrunk the jpegs and added a full res pdf for those that want to print... hope you don't mind... the horizontal scroll is a pain and i'm on dialup at home :-(
All in all, the pre is working but with a few problems. Firstly, upon replacing out the last 380 uF electrolytic cap in the HT PS with a 100uF oiler, ripple rose from 0 to 12mV and it is clearly audible, as it gets in the way of the phono stage.
wow, that sounds like a large increase in ripple for 1/4 the capacitance. simple theory says, if 100uf gives you 12mv, and you double it to 200, you should cut the ripple in 1/2. In any case, 12mv is obviously too much!
That should get cleared up by the HT PS change shown, ie. putting a new nickel(?) choke between the 100 mike and 30 mike caps. While we're at it, might as well change the 80 rectifier to 5ar4, bringing up the voltage, which slightly reduces the value needed for the cathode bypass caps by increasing RL.
i agree that an extra filter section is a good move. not sure i follow on how it will increase Rl, since that is set by the inductance. do be aware that you have 3 parafeed stages here so i would expect it to be very tweeky with cap size and selection. if you do go for a 5AR4, and do not need the extra B+, i would shrink the size of that 25 mic cap to net you the same B+ voltage. the smaller that cap the better in my experiences.
Those caps need to be increased about ten-fold, and I think that is the reason for some midrange raspiness at this point. I hope so.
the bypass caps??? i wouldn't expect increasing their value to help the midrange, eliminating them might though :-)
Also, since the first two plate chokes are only rated for 10ma and are flowing 13.2, I worry about saturation. How would nickel lams change the sound here (assuming new chokes rated at 15ma or so)?
well i'm a bit biased, but i like nickel plate chokes, i wouln't expect the extra 3-4ma to saturate you since your AC signals are so very small in this situation so you are likely trading some AC headroom which you will never use for some extra DC current.
i do find that nickel often removes a grittiness to the sound, but the only way to know if it will work for you is to try it. another source of the harshness you mention is a large palstic cap input filter. your choice of an oiler for the first one is a good one, and getting the size down to a mic or less might also be a big help.

I have no experience with the shunt regulation with the VR tubes, have you given it a listen without them in place??? i know they can be nice little noise generators, and noise seems to be a major cause of harsh sound.
Third, is there a way to try nickel lams in place of the cobalt ones for the output trannies? As shown on the prints, frame height is 1-5/16" with 2 inches between bolt hole centers.
well yes and no... it looks like they are the same size lams as your autoformer so a swap is possible, BUT chances are they are varnished so attempting ot remove the core will surely destroy the lams. *IF* they come apart like the autoformers, you should be able to swap cores for a listen, but i would proceed with caution. as is, they are worth something to someone if you decide to try something different, if you open them up, they become expensive junk box parts.
Last, there just seems to be way too many capacitors in this design. Any ideas on trimming some out without losing sound quality?
i am a big fan of KISS and not putting things into a circuit unless they are needed. the theory behind your design is sound, but sound theory doesn't guarentee sonics. a little time and some rough expeniments will let you know what you can get away with. your biggest friend here will be a variac since it will allow you to try some simple swaps to see if something you thought was helping is really hurting the sound.

if your B+ is on a dedicated transformer you are in great shape, than simply place the primary on a variac and you can set the B+ to whatever you want. the first thing i would look at would be to see what it sounds like withouut the VR tubes. this should be quite simple to try, simply pull the VR tubes and jumper the plate chokes to the B+ then simply dial the variac to keep the B+ the same. ( i really like the lack of decoupling between stages btw)

when it comes to the cathode bypass and ultrapath caps. you really are stuck trying it a bunch of different ways and seeing what happens. I am a big fan of nimh batteries in the cathodes for bias, others like LED's.

the first stage looks like it could do without the cathode bypass or ultrapath caps, using both just seems like asking for trouble. also occasionally remove the teflon bypass caps to see what happens.

lastly I would remove the 10K grid resistor from the 6J5. it is not needed since the autoformer gives a ground reference.

dave
artsybrute
Posts: 6
Joined: Mon Jan 31, 2005 2:57 am

Post by artsybrute »



hey len, i shrunk the jpegs and added a full res pdf for those that want to print... hope you don't mind... the horizontal scroll is a pain and i'm on dialup at home :-(
Thanks, I would have done it myself had I known how.
That should get cleared up by the HT PS change shown, ie. putting a new nickel(?) choke between the 100 mike and 30 mike caps. While we're at it, might as well change the 80 rectifier to 5ar4, bringing up the voltage, which slightly reduces the value needed for the cathode bypass caps by increasing RL.

i agree that an extra filter section is a good move. not sure i follow on how it will increase Rl, since that is set by the inductance.
Maybe I have something wrong. I figured that by going to the 5ar4, I now need more resistance above the plates (plate load?). Using the 417A as an example, we go from 2285+1800 plus the plate DCR of 2075 equals 6160 ohms with the 80 rectifier. Add this to Ra of 1700 at this op point gives 7860. Divide by mu of 44 gives 178.6 ohms for Rk. Put this in parallel with the 250 ohm cathode resistor gives 104 ohms. Then to figure the bypass cap, (1/ 2 * Pi * cutoff frequency * r) becomes 1/ 2* Pi * 1 hz * 104 gives a cap value of 1527uF. If we change to the 5ar4, resistance increases from 2285 to 4830, bringing cap size down to 1310 uF. No?


do be aware that you have 3 parafeed stages here so i would expect it to be very tweeky with cap size and selection. if you do go for a 5AR4, and do not need the extra B+, i would shrink the size of that 25 mic cap to net you the same B+ voltage. the smaller that cap the better in my experiences.
Thanks. I was unaware of that and will try it.
Those caps need to be increased about ten-fold, and I think that is the reason for some midrange raspiness at this point. I hope so.


the bypass caps??? i wouldn't expect increasing their value to help the midrange, eliminating them might though :-)


Also, since the first two plate chokes are only rated for 10ma and are flowing 13.2, I worry about saturation. How would nickel lams change the sound here (assuming new chokes rated at 15ma or so)?


well i'm a bit biased, but i like nickel plate chokes, i wouln't expect the extra 3-4ma to saturate you since your AC signals are so very small in this situation so you are likely trading some AC headroom which you will never use for some extra DC current.
i do find that nickel often removes a grittiness to the sound, but the only way to know if it will work for you is to try it. another source of the harshness you mention is a large palstic cap input filter. your choice of an oiler for the first one is a good one, and getting the size down to a mic or less might also be a big help.
Thanks, will try it. I have a bunch of oilers here.

I have no experience with the shunt regulation with the VR tubes, have you given it a listen without them in place??? i know they can be nice little noise generators, and noise seems to be a major cause of harsh sound.
I'll try that last, as I'm emotionally attached to them :-/
Third, is there a way to try nickel lams in place of the cobalt ones for the output trannies? As shown on the prints, frame height is 1-5/16" with 2 inches between bolt hole centers.

well yes and no... it looks like they are the same size lams as your autoformer so a swap is possible, BUT chances are they are varnished so attempting ot remove the core will surely destroy the lams. *IF* they come apart like the autoformers, you should be able to swap cores for a listen, but i would proceed with caution. as is, they are worth something to someone if you decide to try something different, if you open them up, they become expensive junk box parts.
Thanks for the warning.
Last, there just seems to be way too many capacitors in this design. Any ideas on trimming some out without losing sound quality?

i am a big fan of KISS and not putting things into a circuit unless they are needed. the theory behind your design is sound, but sound theory doesn't guarentee sonics. a little time and some rough expeniments will let you know what you can get away with. your biggest friend here will be a variac since it will allow you to try some simple swaps to see if something you thought was helping is really hurting the sound.

if your B+ is on a dedicated transformer you are in great shape, than simply place the primary on a variac and you can set the B+ to whatever you want. the first thing i would look at would be to see what it sounds like withouut the VR tubes. this should be quite simple to try, simply pull the VR tubes and jumper the plate chokes to the B+ then simply dial the variac to keep the B+ the same. ( i really like the lack of decoupling between stages btw)

when it comes to the cathode bypass and ultrapath caps. you really are stuck trying it a bunch of different ways and seeing what happens. I am a big fan of nimh batteries in the cathodes for bias, others like LED's.
I have a bunch of 3.3V led's that I picked up just for that reason, but then read that some find them very noisy. I'll try them now. If I understand correctly, led's do not need cathode bypass caps, so i should pull the caps out. Is this correct?

the first stage looks like it could do without the cathode bypass or ultrapath caps, using both just seems like asking for trouble. also occasionally remove the teflon bypass caps to see what happens.

lastly I would remove the 10K grid resistor from the 6J5. it is not needed since the autoformer gives a ground reference.
I need the 10k in order to cut down on coupling cap size. I used C = L/R^2. I measured 8 henries at full attenuation (I may be wrong). Without the 10K load, I used 1000 ohms since that is the approximate resistance of the 6C45Pi (thanks PJ). That gives C = 8/1000^2 = 8uF. Putting in the 10K resistor brings the needed cap size down to 1uF. I had the 4.7 Zen and it physically shortened the signal path compared to an oiler, so I threw it in and got back all my missing bass.

Comments?
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

Maybe I have something wrong. I figured that by going to the 5ar4, I now need more resistance above the plates (plate load?).
your VR tubes are setting the B+ and decoupling each stage, so any resistance "above" them shouldn't change things. by all means go for the 5AR4 then use the extra B+ to reduce the first cap in the CLC filter.

i'm not clear on the origins of your formulas for the remaining calculations so a reference would be helpful.

I'll try that last, as I'm emotionally attached to them :-/
try that first... its quick, easy and easily reversible.
n order to cut down on coupling cap size. I used C = L/R^2. I measured 8 henries at full attenuation (I may be wrong). Without the 10K load, I used 1000 ohms since that is the approximate resistance of the 6C45Pi (thanks PJ). That gives C = 8/1000^2 = 8uF. Putting in the 10K resistor brings the needed cap size down to 1uF. I had the 4.7 Zen and it physically shortened the signal path compared to an oiler, so I threw it in and got back all my missing bass.
again i need the source for your calculations... it seems like a ROT that may have gone bad for your particular situation. Rules of Thumb are designed to assure the average person won't screw up too badly. I prefer taking a simpler approach and learning from your mistakes.

dave
artsybrute
Posts: 6
Joined: Mon Jan 31, 2005 2:57 am

Post by artsybrute »

dave slagle wrote:
Maybe I have something wrong. I figured that by going to the 5ar4, I now need more resistance above the plates (plate load?).
your VR tubes are setting the B+ and decoupling each stage, so any resistance "above" them shouldn't change things. by all means go for the 5AR4 then use the extra B+ to reduce the first cap in the CLC filter.

i'm not clear on the origins of your formulas for the remaining calculations so a reference would be helpful.

Not to open a can of worms, but I used PJ's input from this link: http://db.audioasylum.com/cgi/m.mpl?for ... ad&n=90071


I'll try that last, as I'm emotionally attached to them :-/
try that first... its quick, easy and easily reversible.

okay, time to fire up the variac.

n order to cut down on coupling cap size. I used C = L/R^2. I measured 8 henries at full attenuation (I may be wrong). Without the 10K load, I used 1000 ohms since that is the approximate resistance of the 6C45Pi (thanks PJ). That gives C = 8/1000^2 = 8uF. Putting in the 10K resistor brings the needed cap size down to 1uF. I had the 4.7 Zen and it physically shortened the signal path compared to an oiler, so I threw it in and got back all my missing bass.
again i need the source for your calculations... it seems like a ROT that may have gone bad for your particular situation. Rules of Thumb are designed to assure the average person won't screw up too badly. I prefer taking a simpler approach and learning from your mistakes.


BTW, the LED's are in and sound clearer than the resistor/cap combo. Bass is tighter and less prominent. It's a keeper. Still lots of hash though, and only when a record is playing. Even with the turntable running but the needle up, there are no clicks, pops or static that I would associate with HF from the tube heaters.
artsybrute
Posts: 6
Joined: Mon Jan 31, 2005 2:57 am

Post by artsybrute »

Oh, sorry, forgot.

The calculations for the cathode bypass caps are from Morgan Jones' book "Valve Amplifiers" pages 66-67. As I understand it, of course.

I've got all this stuff on spreadsheets in case you want to look at it.


Thanks again,

Len
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

Not to open a can of worms, but I used PJ's input from this link: http://db.audioasylum.com/cgi/m.mpl?for ... ad&n=90071
PJ's ROT's are up there with the best... i see how the 10K tames the resonant hump (see attached images) luckily your inductance (if you have it stacked 1X1 is around 160 hy, and the Rp of the tube is 1500 ohms ot 2K at your op point. the two plots represent a 10K and a 1 meg load... try both and see. I used an excel program from a friend to model it, i'll see if he minds if i distribute it. (he did have some concerns about errors)
BTW, the LED's are in and sound clearer than the resistor/cap combo. Bass is tighter and less prominent. It's a keeper.
one stage or all three??? i often find if you mix it up things sound more natural... (three stages of LEDs can be too much.) the NIMH batteries are also worth a go. (how far apart are the tubes for the L & R channels?)
Even with the turntable running but the needle up, there are no clicks, pops or static that I would associate with HF from the tube heaters.
you lost me on that one...

dave
Attachments
the 1 mic cap is the dark blue, 2 mic magenta
the 1 mic cap is the dark blue, 2 mic magenta
10K.gif (9.29 KiB) Viewed 14314 times
the 1 mic cap is the dark blue, 2 mic magneta
the 1 mic cap is the dark blue, 2 mic magneta
1meg.gif (8.13 KiB) Viewed 14315 times
artsybrute
Posts: 6
Joined: Mon Jan 31, 2005 2:57 am

Post by artsybrute »

PJ's ROT's are up there with the best... i see how the 10K tames the resonant hump (see attached images) luckily your inductance (if you have it stacked 1X1 is around 160 hy, and the Rp of the tube is 1500 ohms ot 2K at your op point. the two plots represent a 10K and a 1 meg load... try both and see. I used an excel program from a friend to model it, i'll see if he minds if i distribute it. (he did have some concerns about errors)
That is a nice spreadsheet to have. Are you recommending that I try a 1 meg resistor in place of the 10K for a trial?
one stage or all three??? i often find if you mix it up things sound more natural... (three stages of LEDs can be too much.) the NIMH batteries are also worth a go. (how far apart are the tubes for the L & R channels?)
The first two stages (phono only at this point). I'll be sure to try the NIMH"s also pretty soon. The tubes for the channels are at least 8 inches apart.
you lost me on that one...
Sorry, just a conglomeration of thoughts. One, that the 417A's and 6C45Pi's are known for HF hash and two, that the raspiness I'm hearing probably is not due to that problem.
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

That is a nice spreadsheet to have. Are you recommending that I try a 1 meg resistor in place of the 10K for a trial?
I chose the 1 meg value to suggest a worst case if the 10K resistor was removed. It also points out what the 10K resistor was doing. I think it would be worthwhile for a few reasons to give it a try without the
10K in place.

Just because excel says the huge resonant hump is there, doesn't mean it really will be there. I dislike the idea of any peak in the sub 20hz area, they can be excited and watch out if they are. A 1V 2hz signal (turntable rumble?) has the same flux as a 10V 20hz signal. Its not easy to measure below 10hz, but it might be worth it try. Just to see if the spreadsheet matches reality.

The first two stages (phono only at this point). I'll be sure to try the NIMH"s also pretty soon. The tubes for the channels are at least 8 inches apart.
8 inches... Well within the reach of a clip lead :-) another quick and reversible thing to try would be to try a shared cathode.

Sorry, just a conglomeration of thoughts. One, that the 417A's and 6C45Pi's are known for HF hash and two, that the raspiness I'm hearing probably is not due to that problem
Its funny, the main reason many say the high Gm tubes sound harsh is oscillation, and I agree if a tube is oscillating at a few hundred khz it does add a harsh edge to the sound. What people seem to forget is all of the steps that they go through to prevent the oscillation (grid stoppers ferrite beads, heater decoupling caps) can also make the sound harsh.

I'll never forget the day I bypassed a ferrite bead on the grid of one with a clip lead. What a difference, the sound just opened up I hate to think of the circuit I made, but from that day on I have built without anything on the grids and have never looked back.
artsybrute
Posts: 6
Joined: Mon Jan 31, 2005 2:57 am

Post by artsybrute »

Just because excel says the huge resonant hump is there, doesn't mean it really will be there. I dislike the idea of any peak in the sub 20hz area, they can be excited and watch out if they are. A 1V 2hz signal (turntable rumble?) has the same flux as a 10V 20hz signal. Its not easy to measure below 10hz, but it might be worth a try. Just to see if the spreadsheet matches reality.
Oh, it was definitely there. Changing the cap from .22uF to 14uF smoothed the response right out. Then adding the 10K resistor and changing the cap to a 4.7uF Zen made the sound a tad less clear but more musical than the 14uF Unlytics.
8 inches... Well within the reach of a clip lead :-) another quick and reversible thing to try would be to try a shared cathode.
What's a shared cathode?

Its funny, the main reason many say the high Gm tubes sound harsh is oscillation, and I agree if a tube is oscillating at a few hundred khz it does add a harsh edge to the sound. What people seem to forget is all of the steps that they go through to prevent the oscillation (grid stoppers ferrite beads, heater decoupling caps) can also make the sound harsh.

I'll never forget the day I bypassed a ferrite bead on the grid of one with a clip lead. What a difference, the sound just opened up I hate to think of the circuit I made, but from that day on I have built without anything on the grids and have never looked back.
I don't like having all that junk in there either. As soon as I find the cause for the harshness, those parts will come out one at a time if they are found not to improve the sound.

The HT PS is being changed to LCLC at a lower voltage, which will also allow simply disconnecting the VR tubes as a test without using the variac. If neither of those proves helpful, I'll try Morgan Jones' power supply mods that cut down on just such noise.
Post Reply