Power supplies and smaller p2p boards for 2ch tube preamp

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

My3gger

Well-known member
Joined
Nov 2, 2012
Messages
568
Location
EU
Hi,

i'm making another 2ch tube preamp (Redd 47), this time it is for really talented singer who benefits from such preamp, the other channel will be for different sources. I find this quite rare as most musicians only get better sound in the studio, if at all. Her voice also live, where letters in high frequencies like s, c and similar are finally as sang, beside other advantages like tight bass, a lot more details, much more natural sound, etc. It was very surprising hearing such difference for better.
Group has at least two concerts per week, often four, two per day if acoustic... Gear is stored in place with temperature from ~ -3 degrees C to + 30 degrees C, used outside live it can be in ambient from -10 degrees C to ~ +40 degrees C. This is big challenge for me, of course it takes a lot of thought getting it inside standard rack, allow for modifications, turning transformers, getting  as needed in general (made many preamps myself), etc. I love it a lot anyway, sound is so much better after test on a concert and comparisons to mixer.

PSU is first thing i need to make, properly and small, hope to manage dual mono with one power trafo, i see no reason for separating phantom. Requirments are 290V/18mA HT, 6,3V/500mA heaters and 48V/10mA phantom per 1ch.
For HT i used TL783 as Gyraf's G9, Maida with LM317, capacitance multiplier with bipolar transistor-zeners into base and only a few times RC filters. Maybe one or two other similar with this.

1. TL783 needs cooling with heatsink, two quite big ~150u caps where smallest are snap in (not too useful for veroboard), datasheet show quite a few precautions against desctruction, simple discrete design instead IC can be advantage, minimum current for regulation might not be a problem with proper resistor.
http://www.boemboemklap.nl/diy/b_plus_psu.gif

2. Maida is more complicated, needs heatsink and again several precautions against shorts. Smaller components as advantage here because of space. I don't have much experience with it, beside making it for much less current than 18mA at similar HV. Datasheet: http://www.ti.com/lit/an/snoa648/snoa648.pdf
https://i.stack.imgur.com/pBaHV.jpg

3. Capacitance multiplier with bipolar tr. works well every time and has noise in uV, without heatsink. Attachment show it without cap values, normally they are 150u input, the rest 47u. Judging by simple RC filters and compare to cap multi it has two RC filters, not quite enough for what i would like, can output 47u filter enough to get me into low enough range to power one channel? Can't remember if input and output cap were both 150u and others 47u for 290V/18mA.
http://static.flickr.com/139/324693062_1bcc2f4ae2.jpg

4. Capacitance multiplier with IRF840 instead of TIP50 as suggested by TV. Circuit should be pretty small and simple, without heatsink, i don't expect output shorts, although it would be nice if it would survive them for short periods like 30 seconds max as TIP50 one does. I tried hard but couldn't calculate even aproximate gm of FET to find out myself how much it magnifies capacitance and pick right cap for the job. This are TV's propositions:
-Each R/C section (there are two of them - R1C2, R2C3) should have "corner" frequency of cca 1-2Hz. The "gate stopper" could be 330R-1K. Add the gate protection zener as well.
-You will have to calculate the first RC section with the Zener current in mind, i.e. the R1 will have to pass the proper current to the HV zener. R1 will have to be a 2W type.
-In theory, if you use 56K resistors, you could use 2,2uF for the C3 and 22uF for C2 (or possibly C4). C1 and C4 you will have to come up with your values.

5. PRR suggests simple RC filters for projects like this, says it takes about 1uF per mA and tells it is time proven concept. It probably means each such cap in any of three filters because 22u seems pretty low. So looking at classic like V72 it has 8uF after gretz, the rest are two RC filters at 2k5R/8uF.
V76 with gretz into 32uF two RC filters at 8kR/2x32uF-8k/32uF, HT current is 20,7mA, not that far away considering German engineers.
This is very simple, higher current output tube is taken from first RC filter at 8kR/2x32uF and 14,6mA/300V, the rest goes into other 3 EF804 tubes with total 6,1mA current and lower than 300V . I like this approach, 47uF caps are small, high quality and cheap, circuit can be small without big resistors. Ian's psu confuses me with RC filters consisting from 220uF caps for one channel and lower current than V76 or Redd47. Might be designed so to power more channels, can't find other reasons.


Heater and phantom are easy, small and cheap, so no problem to add them. It takes heatsink for something like LT1085 LDO regulator for heaters. I could use one with two regulators because each channel takes only 500mA, plus some extra for relay if needed, LEDs are in mA range.
Tried several drawings to get it into 2U rack, it doesn't seem possible with 4 tubes. Another thing to try is having two L profiles each for 2 tubes and two turret boards because they allow getting closer to tube socket and making current loop small and properly wired. This way secondary wires of IT into grids can be very short, front panel for wiring switches is close, output trafos can be between profiles. Tubes will need covers because of moving, despite tight pin Micalex sockets. Those take ~20mm more to get them out if put horizontally. I will most likely have to take max depth 3U, so i don't run out of space as it happened once with input trafo...


A lot of abuse is expected because of moving, compared to Chandler good build will last a lot longer, won't need all those extras to distract soundguy. I was able to find good ergonomics with several revisions, i'm sure original engineers designed for this too, instead of fader it will have output pot. One option is Bourns T-pad, i have two old back ones left, dual pot seems to interfere with the signal too much, also have a lot of mismatch in log curve. Short runs with 1k ohms pot ala API with two small value resistors (NYD wrote schematic) in front seems best option. Have to check it some more before deciding.

I know this is a lot of writing and ideas, if someone feels it would be better splitting between different posts i will do it sooner or later. Space between sections is left because for easier reading and no one will reply to all of it, feel free to respond with thoughts or experiences about any part. Similar examples are very welcome, i could only find some without OT, or other formats, builds meant for not much moving, etc.
As mentioned it is important challenge for me, so thanks for any comment.





 
Thank you Ian, first thing done was checking your page. Power section didn't show up for some reason, it does now.  Wish i could use Eurochannel here, for sure it will be used for next 2 mics and other things Redd47 can't do.
Now i understand why high value caps, HT 250 will be just fine. It is not yet clear to me if 120dB of ripple attenuation is achieved at 15mA, while over 80dB at 100mA, not counting global NFB in preamp. I found TH caps of similar value and quality, so it is just matter of calculating correct resistors for needed voltage. Board for dual mono can be even smaller this way, can't put pcbs into such chasis.
I didn't find good schematic for DC heaters with RCs and can't experiment much here. At 500mA heater per channel (dual mono for several reasons) it is very simple to use new regulators like LT1085 and schottky diodes, works for me every time even with cheaper LM1085. It is interesting to notice that ripple rejection with LT1085 is ~90dB, very close to worst case with your HT250.... I will need DC heaters to power a few leds, maybe relay, who knows what else with time. Pentode also has quite high gain.
I remember John Hinson mentioning how regulation is part of the sound of Redd47 which is supposed to be written in original datasheet. Can't see that and those OA2s are no regulators but voltage references like zeners, mostly causing problem and no one knows how long they will be available. Difference in sound might be because we think it is, at least i do when making 1ch and playing with different references, regulators and such things. Bigger they are, "better" is sound and then i tend to comment how good it is... Could be psychologic and not so much electrical phenomenom.
RC filters seem to be the right and safe answer here, so thanks for helping decide with confidence.

p.s.: Phantom will be TL783 from your schematic because resistor takes care of minimum current.
 
My3gger said:
Thank you Ian, first thing done was checking your page. Power section didn't show up for some reason, it does now.  Wish i could use Eurochannel here, for sure it will be used for next 2 mics and other things Redd47 can't do.
Now i understand why high value caps, HT 250 will be just fine. It is not yet clear to me if 120dB of ripple attenuation is achieved at 15mA, while over 80dB at 100mA, not counting global NFB in preamp. I found TH caps of similar value and quality, so it is just matter of calculating correct resistors for needed voltage. Board for dual mono can be even smaller this way, can't put pcbs into such chassis.
Ripple attenuation of each stage is just 2*PI*f*R*C where R and C are the resistor and capacitor values and f is the ripple frequency. For 1K and 220uF and 100Hz this gives 138 times or  42.8dB per stage.

Peak to peak ripple voltage is given by V =i*t/C where i is load current, t is ripple period (10mS for 100Hz) and C is the capacitance. For a 15mA load this gives a ripple of 0.68V peak to peak which after 120dB of attenuation is less than 1uV.

For 100mA operation, the resistors need to be reduced to 200 ohms so the attenuation per stage becomes 27.6 or 28.8dB per stage which is 86dB for three stage. Ripple voltage at 100mA is 4.54V. After 86dB of attenuation this is 227uV peak to peak or about 78uV rms
I didn't find good schematic for DC heaters with RCs and can't experiment much here. At 500mA heater per channel (dual mono for several reasons) it is very simple to use new regulators like LT1085 and schottky diodes, works for me every time even with cheaper LM1085. It is interesting to notice that ripple rejection with LT1085 is ~90dB, very close to worst case with your HT250.... I will need DC heaters to power a few leds, maybe relay, who knows what else with time. Pentode also has quite high gain.
I also like the LT1083/4/5 for dc heaters. If you look in the lunchbox folder under the DIY tab of my web site you will find the lunchbox PSU designs which include dc heaters.

There are some simple RC designs for heaters; I think Alec used it in one of their mixers many many years ago.

Cheers

Ian
 
ruffrecords said:
Ripple attenuation of each stage is just 2*PI*f*R*C where R and C are the resistor and capacitor values and f is the ripple frequency. For 1K and 220uF and 100Hz this gives 138 times or  42.8dB per stage.

Peak to peak ripple voltage is given by V =i*t/C where i is load current, t is ripple period (10mS for 100Hz) and C is the capacitance. For a 15mA load this gives a ripple of 0.68V peak to peak which after 120dB of attenuation is less than 1uV.

For 100mA operation, the resistors need to be reduced to 200 ohms so the attenuation per stage becomes 27.6 or 28.8dB per stage which is 86dB for three stage. Ripple voltage at 100mA is 4.54V. After 86dB of attenuation this is 227uV peak to peak or about 78uV rmsI also like the LT1083/4/5 for dc heaters. If you look in the lunchbox folder under the DIY tab of my web site you will find the lunchbox PSU designs which include dc heaters.

There are some simple RC designs for heaters; I think Alec used it in one of their mixers many many years ago.

Cheers

Ian

Thanks for confirming my homework, 220u will be needed here compared to 150u, if i want over 100dB of ripple attenuation at 18,5mA HT. This results seem better than cap multipliers with TIP50, can't say for sure because ripple is so low that scope doesn't really show exact quantity, although it is good for confirming visualy.

You use snap in 220u/400V 25x35mm where miniaturisation isn't mentioned:
http://uk.farnell.com/multicomp/mcklz400m221n35y/cap-alu-elec-220uf-400v-snap-in/dp/2610784
My choice might be through hole 220u/450V 18x50mm, miniaturisation mentioned:
http://uk.farnell.com/rubycon/450qxw220mefc18x50/cap-alu-elec-220uf-450v-rad/dp/2342120
It has very similar data to yours and 450V max, max current shouldn't cause problems in this application.
I was told bigger caps might last longer, maybe because of heat which is managed very well here anyway.
Do you think miniaturisation could influence performance of cap, or it might also last long time if used properly? TH will be needed here, this is why i ask.

My supplies mostly have two trafo taps for heaters just in case (custom is very cheap in eastern Europe), heatsinks are cut in half and each screwed into veroboard, so LT/LM108x regulators get no more than ~60-70 degrees C, counting ambient temperature at 500mA/6V like here. It seems heat is a problem, datasheets show max allowed conditions and many think one diy device can be compared to series where many revisions were made. Many times it is run to extremes and fails.
 
Regular leaded electrolytics should be fine. As you say, heat is not a problem and the ones you picked are rated for 105 degrees.

LT1084 at 500mA should be no problem but it sounds like your heatsinks might be a little on the small side. I generally aim for no more than a 20 degree rise of the heatsink. You can get an insulating kit with a mica washer that allows you to screw them direct to a chassis.

Cheers

Ian
 
Thanks, i will follow your advice regarding rise of heat. A few heatsinks i cut came from old TV's thrown away, they had much better dissipation than many bigger. Fins are thinner with more area and this is why rise isn't very high, although idea of getting no more than 60 degrees C at high ambient temperature is probably what makes the difference between long lasting regulator, or fails like many mention.
 
I build my tube preamps into small aluminium cassettes that plug into a eurorack. They are 3U high, 2.8 inches wide and 160mm deep. They are fully enclosed (no holes) and have three tubes inside which dissipate a total of about 10 watts. You might expect these to get pretty hot but there is so much total metal dissipating the heat that the don't. They rise by 20 degrees at the most. The front panel feels warm to the touch but that was the case with many of the Neve class A consoles made in the 70s.

Cheers

Ian
 
I always use four 1N5822 schottky diodes as a bridge used for heaters:
http://www.vishay.com/docs/88526/1n5820.pdf

Iirc they get barely warm at 500-700mA and 6V, since this doesn't sound logical to me i will get a probe this days and recheck. Could be something else than power heating them. There are problems in this application because of 1,3mm pins which need holes drilled into veroboard, then it is a little hard to bend them as needed. "SIL" rectifiers as in Lunch Box also need bigger holes which would make them move with time, i'm trying to make this small but still reliable.
Would something like this work instead, schottkys are not really needed:
http://si.farnell.com/vishay/w01g-e4-51/bridge-rectifier-1-5a-100v/dp/1497575
It is 1,5A 100V bridge rectifier, i calculated 0.35W of power on it which should dissipate well. Am i correct about it?

HT will have this one:
http://si.farnell.com/vishay/w10g-e4-51/bridge-rectifier-1-5a-1000v/dp/1497580

And phantom this:
http://si.farnell.com/vishay/w01g-e4-51/bridge-rectifier-1-5a-100v/dp/1497575

Last two seem ok to me, drawing from second revision is how i want psu. It must go to pcb for next projects for obvious reasons, amp can still easily be p2p if i want it this way. Veroboards are sometimes low quality and break easily by drilling, maybe Farnell sells higher quality ones like Banzai had.


 
I rarely see a warm HT or phantom rectifier. I do see hot heater bridges but only when delivering 3 amps or so. Below an amp they should be just warm.

Cheers

Ian
 
I'm more or less done with power supplies, thanks a lot Ian.
Now it is time to choose boards with pins or turrets, not too many questions left. Preamp circuits with proper components on veroboard were fine on some projects. Components can be radial or axial, smaller, better and shorter current loops, less wire crossing, etc. I'm not sure even smaller turrets will satisfy regarding layout, they take a bit more space. It is already crowded with two supplies,  bigger sig. transformers than most builds here. It would also be nice to get tubes with shields horizontaly on L bracket..
The only problem were tiny bit too short pins on solder side, i got probably less than 1mm to solder, so two pins from about 20 don't hold well after about 5 years. This are actual pins used before:
http://www.tme.eu/en/details/1364d.68/solder-terminals-pcb-mount/vogt/
Tip on soldering side looks more like this:
http://www.banzaimusic.com/Solder-terminals-1-3x11mm-100-pieces.html

I found 1mm pins with 3mm lenght below flange, 15,5mm lenght, other 12mm:
http://si.farnell.com/ettinger/13-14-145/terminal-turret-1mm-non-insul/dp/2691436
http://si.farnell.com/ettinger/13-14-125/terminal-turret-1mm-non-insul/dp/2691437
I'm not really sure veroboard with 1mm holes will take 1mm pins, guess they should. 12mm pins might hold better when soldered because they don't have tip.

There are also differences in quality of boards, epoxy fibreglass or epoxy glass composite with square 2,2mmx2,2mm pads instead of rounded seem to be better choice.
This is veroboard that seems right to me, better than those before and would allow building 2ch with some cutting:
http://si.farnell.com/roth-elektronik/re210-s1/pcb-prototype-re210-s1/dp/1221149




 
I did, but it would most likely take too much space for all i need. Maybe i'm a bit too picky with layouts because i want them similar as is schematic, so no decoupling just anywhere, it starts with input and goes to output with hefty ground wires. Things like HT are separated, or other parts with possible interferences are away from sensitive points like input sec, input grid, so it is made quite similar to V72/76 where HT comes into circuit from below. Chosen material are axial and radial components, a few resistors already need to be pointed upwards as you have them.
Difference is maybe 20mm per ch, it adds up with 2ch, 2 supplies, horizontal tubes with shield and xlr connectors behind. Front plate here also takes some more space because of NFB switch, this is another ~20mm. Can't say if chasis maker will put xlr holes so that they won't interfere with positions of tubes and OTs. I have to allow for possible modifications, although it will most likely be as it is, with more controls.
One of this preamps was not very quiet because of layout on tag board, simpler and smaller circuit of RCA BA-2 is just perfect this way. I might be complicating about layout and have to try different approaches, it is much easier with 1ch and more time because i have a lot more other work now. But will check your pcb layouts because they can give answers to many questions.
I've seen many builds with interferences because of bad p2p, mostly more skilled people meassure and post this things.
This is BA-2 example, heater heatsink should be turned the other way around, it didn't need longer chasis. I could barely see AC at several points with HT. It is also very quiet with EH EF86 wired as triodes, not far away from solid state preamps with more gain, better than some.
 

Attachments

  • RCA BA-2 inside.jpg
    RCA BA-2 inside.jpg
    810.5 KB · Views: 109
Front too so people can see the whole thing
 

Attachments

  • RCA BA-2 front.jpg
    RCA BA-2 front.jpg
    614.3 KB · Views: 78
Turret board like this will probably be much better:
http://www.tubeampdoctor.com/en/shop_Amp_Parts_Eyelet_Turret_Boards/TAG_BOARD_257_x_66_7_x_3mm_red_brown_3263
And slotted turrets, i'm waiting for reply about their dimensions because smaller slotted are fine for radial components too:
http://www.tubeampdoctor.com/en/shop_Amp_Parts_Eyelet_Turret_Boards/Turret_round_slotted_version_50pc_package_3276

I had 10% failure with pins, almost don't see them used this way, veroboard might be too thin for longer board and durability. So it is out of question, could be fine for prototyping. Hope to get reply soon and go on with project.
 
I am a little confused. The turret board you linked to is 257mm by 66.7mm. The tag strip I linked to is only 118mm by 38mm yet you said the latter was probably too big.

Cheers

Ian
 
I should explain it better. It is not BA-2 but Redd47 with more components, so 2ch tag strip would become quite big compared to tag board with tags where i need them and 189 holes all together. I like being careful with current loop; make it short, separate IT sec wires from HT and places with more current, input grids, have decoupling (2 this time) where needed, possibility of easily rotate signal trafos and get them away from hot E88CC, have tubes horizontaly because this allows turning them and use shields, etc. I've seen to many 2ch builds where 2nd channel had problems, so why not make layout the way it works every time. I know it is more about positions of transformers, although i can clearly see good layout helps a lot too, it seems especialy fat (uninterupted) grounds where possible and mentioned separation.
So already drilled 3mm board without turrets allows for having parts where i want them.  I will cut it to about 140x50mm if based on latest layout revision, probably more or less to dimensions of the board you linked. I like putting a lot of thought in layouts because it is rewarding to me  as much as right signal transformers. Could happen i would have to use 2 tag boards which is what i like to avoid. Costs a lot more which doesn't bother me too much because components are all good but nothing exotic, again idea from your builds. I need that tool for boards without holes when doing next projects. If you think those small slotted turrets aren't good idea, or i'm complicating with layout or anything else i can still change for better.
 
After checking Ian's phantom supply with TL783 it looks like 2W 4k7 resistor would act as fast ramp down of output voltage. My thinking is 4k7 takes minimum current required by TL783 and output cap isn't so big it would take longer time to empty it.
So this should be ok in cases someone turns off phantom and right away connects something phantom could damage. I had situations myself when a few seconds more of phantom could damage something.
What about slow ramp up as needed for V72 trafos and similar, can't find another reason for it, maybe they are? There is some delay before cap is full, maybe not enough for very tiny windings? Does regulator takes care of this, can't find it in datasheets, i know it is short circuit proof.

I would rather use Ian's plan because everyone until now worked very well and was simple. Used to do all sorts of things for minimum current before idea of 4k7 resistor. Thing is also regulated and has enough filtering.
Schematic:
http://www.ianbell.ukfsn.org/pics/Phantompsu.png

Another phantom schematic i used mostly is from Bo Hansen:
http://www.hansenaudio.se/API%20psu.jpg
I start with 470-1000uF per channel and add ~10uF cap over zener to lower noise. The rest is as on schematic, each channel has at least 1k and 100uF, if not full circuit.
Iirc there is some slight delay before full 48V, enough for anything.
Ramp down takes 1-2 sec which is enough.
Noise is low.
It is short circuit proof.

Can't easily try TL783 psu at the moment, i did Bo's some time ago and yesterday, it comfirmed what i already knew and experienced. I don't mind which one it is as long as it does at least what Bo's does. Regulation with TL783 is simple and better power for mics, this is why i'm asking before sending order.

 
My3gger said:
If you think those small slotted turrets aren't good idea, or i'm complicating with layout or anything else i can still change for better.

I think slotted turrets are fine for p2p. As you say they have the advantage of allowing you to create exctly the layout you want.

Cheers

Ian
 

Latest posts

Back
Top