SRX-Plus Build Thread

[mwl168]

Thanks Kevin!

 

[JimL]

11 hours ago, kevin gilmore said:

because that is how jiml gets away with half the amount of parts.

for the srx circuit it works because a low impedance ground is not necessary.

i myself would not do this, i would do the usual 2 supplies tied together. It only took me about 1 hour to do that layout, so no biggie.

 That's me!  Cheap and cheerful!  

[Blueman2]

On 12/20/2015 at 7:10 PM, mwl168 said:

Apologize for my absence. Have been busy working on a project that's in trouble.

I will contact PCBNET Monday and get some quotes. Giving the tubes are mounted on the board, I suggest we go with 2mm thick and 3oz copper PCB. It'll add to the cost but I think it's worth it.

mwl168,

Any chance of going forward with the GB for just the amp?  I think some people (like myself) might go with another power supply for now.  I got hold of a BH PSU board and plan to start building that very soon.  It would be great to the the SRX-plus boards so I can start that as well.  

Thanks again for be willing to lead the board buy.

Blueman2

[UFN]

1 hour ago, Blueman2 said:

mwl168,

Any chance of going forward with the GB for just the amp?  I think some people (like myself) might go with another power supply for now. 

Maybe take orders for the amp and the PSU separately instead? I'd at least still want a full set but I might consider adding an extra amp board - just in case. That would allow people to order amp boards only as well? Does give mwl168 some addition complexity in packing and shipping boards though.

//UFN

[mwl168]

No problem for me if people prefer to order amp and PS board separately. We just have to keep in mind that small quantity will drive the cost high. 

I plan to finalze the GB next week.

[JimL]

On January 8, 2016 at 5:44 AM, kevin gilmore said:

because that is how jiml gets away with half the amount of parts.

for the srx circuit it works because a low impedance ground is not necessary.

i myself would not do this, i would do the usual 2 supplies tied together. It only took me about 1 hour to do that layout, so no biggie.

More seriously, though, kevin gilmore is correct.  I am a big fan of the Dyna Stereo 70 "school" of design - simple, high performance circuits.  That's what the original SRX was, and that's what the SRX Plus is, including the power supply.  Simplicity has its advantages but a more complex circuit can be higher performance - in fact, what is the justification for a more complex circuit if it is NOT higher performance, other than mental masturbation?  That's what engineering is all about, making choices balancing performance and complexity/cost.  Another way to do a shunt regulator is duplicating the circuit and using one for the positive side and one for the negative side, tying the two together at the outputs.  Likely somewhat better performance, but - twice as many parts.

[Blueman2]

OK, I have assembled all the parts for a KGBH PSU build to go with one of the SRX-Plus AMPs. I plan to build 1 AMP with KGBH PSU and 1 AMP with JimL's shunt PSU.  Just to see the difference (If I can tell any difference!).  One will stay with me, one will go to my son.  

I was asking questions about build in the SRX-Plus thread but decided they belonged better here so as not to pollute the SRX-Plus design thread.  Sorry JimL for taking you down a rabbit hole in that thread!  

Soooo.  I am using the kgbhultraminipsv4 PCB for my KGBH PSU.  There are 2 things I am having to deal with using this PSU for the SRX-Plus.  

1. B+/B- output.  The kgbhultraminipsv4 PSU board says on the silk screen 350V.  But it has zener diodes listed as 100,150,150 on each rail.  That adds to 400V.   Is that a mistake on this board?  I could not find any threads specific to this board talking about this issue. Also, KG's SRX6 board lists rail input of 340V.  Is 350 close enough or should I add a 90V zener in place of one of the 100V.    
2. -20V output.  The SRX-Plus requires -20V for the input stage tail constant current sources.  In JimL's design, he used resistor(s) from the -325V rail, calculated based on the current draw of both the channels, to get to -20V.  My KGBH PSU does not have that and, as mentioned above, will be at -350V on that rail.  I guess my options are:
   1. add a resistor (I think 143K ohms would do) from the -350V rail to create the needed -20V input on the SRX6 board.  JimL, how sensitive is the CCS to exactly 20V?  Or is +/- 10% OK??     
   2. I could use the unneeded +/-15VDC curcuit on the  kgbhultraminipsv4 PSU board, replace the 7915 with a 7918 (there is no 7920 that I can find!) and use the -18V output.   I think -18V is close enough to -20V (??). 

The advantage of using the 143K resistor off -350V rail is it saves a transformer and some devices (7918), so I will probably go that route.  I plan to still use a 7905 or 7912 for +5 or +12V in order to drive power LED and a fan for my case.  

Edited January 14, 2016 by Blueman2

[gepardcv]

About the B+/B- output: the zener strings as specified do produce +/-400V rails. If you want +/-350, change both zener strings to add up to 350. I suggest you actually use a zener string adding up to 340 or slightly less, since zeners tend to run at a higher voltage when warmed up, i.e., if you make them add up to 350, you'll probably end up with 360 or more. I also hand-match zeners so they really do add up to the rated voltage, since they're normally within 5% tolerance, but's probably overkill.

Not sure about the -20V line.

[JimL]

11 hours ago, Blueman2 said:

 

Soooo.  I am using the kgbhultraminipsv4 PCB for my KGBH PSU.  There are 2 things I am having to deal with using this PSU for the SRX-Plus.  

1. -20V output.  The SRX-Plus requires -20V for the input stage tail constant current sources.  In JimL's design, he used resistor(s) from the -325V rail, calculated based on the current draw of both the channels, to get to -20V.  My KGBH PSU does not have that and, as mentioned above, will be at -350V on that rail.  I guess my options are:
   1. add a resistor (I think 143K ohms would do) from the -350V rail to create the needed -20V input on the SRX6 board.  JimL, how sensitive is the CCS to exactly 20V?  Or is +/- 10% OK??     
   2. I could use the unneeded +/-15VDC curcuit on the  kgbhultraminipsv4 PSU board, replace the 7915 with a 7918 (there is no 7920 that I can find!) and use the -18V output.   I think -18V is close enough to -20V (??). 

The advantage of using the 143K resistor off -350V rail is it saves a transformer and some devices (7918), so I will probably go that route.  I plan to still use a 7905 or 7912 for +5 or +12V in order to drive power LED and a fan for my case.  

Actually, one of the advantages of using a cathode tail constant current source (CCS) is that it is very INdependent of the voltage at its far end, as long as you don't exceed the power dissipation of the CCS, so anything from about 10 volts up will work.  I used two resistors because I wanted to bias the output filaments near the cathode voltage of the 6SN7GTA tubes, but if you don't need that (you can just let the filaments float), you can calculate the needed dropping resistor as follows:

 

R (dropping resistor) = [(B- voltage) - 20]/2.6 mA.  If you use the original 300k plate resistors for the inputs, use 2.2 mA instead of  2.6 mA.

 

So, for example, for B- voltage = -350volts, R = [350-20]/.0022 = 150k,  Hmm, I need to make a correction. For my 325 volt build and 250k plate resistors, the total R should be around 125k, not 143k.

Edited January 14, 2016 by JimL

[Blueman2]

Cool.  Thanks Jim.  I was going to put probes on this anyway to reconfirm what is necessary for the 20V, but will start with 150K off the -350 rail.  Yes, I was going to use the 300K plate resistors per the KG board.  

And what is the advantage of not letting the filaments float in your design?  Just curious to learn.  

BTW, you mentioned that you somehow used a 6.3VAC winding to drive the 12AT7 tubes in your original design.  How is that possible?  They need at least 12V don't they? Or did you uplift the voltage somehow using diodes?  Just curious to see if I can still use my 2x 6.3VAC windings to drive everything, or if I will just bite the bullet and get a 12.6V transformer too.  

Edited January 14, 2016 by Blueman2

[JimL]

7 hours ago, Blueman2 said:

 And what is the advantage of not letting the filaments float in your design?  Just curious to learn.  

BTW, you mentioned that you somehow used a 6.3VAC winding to drive the 12AT7 tubes in your original design.  How is that possible?  They need at least 12V don't they? Or did you uplift the voltage somehow using diodes?  Just curious to see if I can still use my 2x 6.3VAC windings to drive everything, or if I will just bit the bullet and get a 12.6V transformer too.  

So the reason for fixing the base voltage of the filaments by using a resistor dividing string is to make sure the cathode to filament voltage is well within safe limits.  Doesn't need to be done - the original SRX just had two filament windings and let them float.

 

All 12A_7 heater filaments can be run off 6.3 volt or 12.6 volt  depending on how they are wired, because there are three pins that are connected to the filament, with one of the pins (pin 9) connected to the midway point on the heater coil. Get a tube manual and look at the base connections.  For 12.6 volts, the heater voltage is connected to pins 4 and 5, for 6.3 volts, pins 4 and 5 are connected together and then the heater voltage is connected between those pins and pin 9.  The srx6 board has the heater connections for the 12AT7s running only to the 12.6 volt pins, to prevent the builder from trying to cheap out and use the same winding to power both output and input filaments.  If they are different voltages they have to be different windings.  KG was trying to make it as idiot-proof as possible, always a good idea when playing around high voltages.

[Blueman2]

Thanks again.  I was wondering why 9 was not connected to anything.  I think I will do some soldering on the board to add a connection to Pin 9, and just use my other 6.3V winding.   Now I am back down to just 1 transformer.  Yeah!

 

Any recommendations for the Volume Control that I assume goes before the inputs? I know one can drop a lot of cash on these attenuators, so just looking for something that is in line with the cost of this overall build.  Is something like this acceptable??

Stepped Attenuator Log 100K

 

Edited January 14, 2016 by Blueman2

[JimL]

Well be careful if you're using the srx6 board.  If you just connect pins 4 and 5 you're shorting the filament winding because it is connected between pin 4 and pin 5 - power up and instant transformer burn out.  You would have to break the connection to either pin 4 or pin 5, run the broken connection to pin 9 and connect pins 4 and 5 with a jumper.  

 

No recs on volume control, except that Alps RK27 doesn't seem to be well regarded.  Goldpoint stepped attenuators and TDK pots seem to be well regarded but pricy.  Alps RK50 is the gold standard but you need lots of gold for it, and it's shaft also has a bigger diameter than the usual 1/4" so you need a custom knob.  spritzer likes the Alpha pot but don't know where you can get it in small quantities and I don't think he is selling them any more.

[Blueman2]

14 hours ago, JimL said:

Well be careful if you're using the srx6 board.  If you just connect pins 4 and 5 you're shorting the filament winding because it is connected between pin 4 and pin 5 - power up and instant transformer burn out.  You would have to break the connection to either pin 4 or pin 5, run the broken connection to pin 9 and connect pins 4 and 5 with a jumper.  

Just to make sure I have this right, I was planning to CUT the trace going to connector 2 labeled "Filament" on the SRX6 board. Then run a wire from each pin 9 of the 12AT7s, and run that to connector 2.  Then I would also have to short one of the 4-5 pins of one of the tubes so that all tubes would have 4 and 5 shorted together (in the diagram below, I shorted 4/5 on the lower left tube).  So pin 9 gets one side of the 6.3V, and pins 4/5 both go to the other side.  That right?  

Note: for the 12AT7s, I am using the normal tube numbering for the pins, so ignore the numbering in the SRX6 layout below.  Not sure why they are different.  So pin 1 is A1, then clockwise from there.  

Edited January 15, 2016 by Blueman2

[JimL]

That should work.  Note that the tube sockets are mounted on the opposite side of the board from the other components (resistors, caps, etc.) so you'll have to route the wiring around them, assuming you're going to solder to the bottom of the sockets.

[Blueman2]

6 hours ago, JimL said:

That should work.  Note that the tube sockets are mounted on the opposite side of the board from the other components (resistors, caps, etc.) so you'll have to route the wiring around them, assuming you're going to solder to the bottom of the sockets.

OK, just to show how much of a noob I am, I did not know that the tubes were mounted opposite to the components!  Is that to allow the tubes to fully penetrate a case top?  Or to prevent the heat of the tubes from impacting nearby components?  Or both?  I will need to rethink my strategy for the case for this system.  I was going to repurpose another case I already had, but this might change things.  I will need to hunt around for pictures of how others have done this.  Maybe mount this onto the TOP of my case rather than the bottom, and use a bit of extra wire to the inputs, outputs, and power supply to allow for the top to be removed.  I need to research on how others have done this. 

Edited January 16, 2016 by Blueman2

[JimL]

Possibly to make it easy to replace tubes, or as a style choice.  The tubes don't make all that much heat, unlike EL34s in the DIY T2, BHSE, Megatron, etc.  In my build they were inside the case, and the KGST, which makes a comparable amount of heat, also has the tubes inside the case.  In this case it's just what Dr. Gilmore decided to do.  See pongo5's build in the SRX thread for an example of how to build it.  

[Blueman2]

I added a Bill of Materials to the first post for the SRX-Plus amp.  Anyone else who is looking to build, please let me know if you see any errors or omissions.  I will keep the first post updated with the correct BOM.  

[mwl168]

51 minutes ago, Blueman2 said:

I added a Bill of Materials to the first post for the SRX-Plus amp.  Anyone else who is looking to build, please let me know if you see any errors or omissions.  I will keep the first post updated with the correct BOM.  

The board is set up for TO220 package DN2540 only. The TO92 package has different pinout.

[Blueman2]

21 hours ago, mwl168 said:

The board is set up for TO220 package DN2540 only. The TO92 package has different pinout.

Fixed.  

 

[Blueman2]

Dr. Gilmore,

On your board design for the amp (SRX6), you used 0.22uf/1000V caps in place of the original 0.1uf 1000V between the upper 12AT7 and the 6SN7GTA.  Is that a critical change or can I still use 0.1uf caps there?  I already have the 0.1uf caps and was hoping to be able to use them.

 

Reply: Kevin responded by PM saying the increase capacitance might improve low frequency performance.  So I am ordering .22uf caps.  

Edited January 22, 2016 by Blueman2

[JimL]

BTW, let me emphasize that it is absolutely ESSENTIAL that the heatsinks for the PS board are attached to the MOSFETs before it is tested, otherwise they can and probably will blow up.

[MLA]

On 14 januari 2016 at 2:49 PM, JimL said:

So the reason for fixing the base voltage of the filaments by using a resistor dividing string is to make sure the cathode to filament voltage is well within safe limits.

So, what base voltage would you recommend, and how to set this up on the SRX Plus board? (or maybe the input tubes are well within cathode-filament voltage limits already)?

[JimL]

The input tube cathodes are sitting at around 0 and +80 volts give or take so I set the input filament tap at around +60.  The output filaments are sitting at around -310 volts (for a -325 volt PS) so I set the output filaments around that voltage.  These filament "bias" voltages are incorporated into the shunt regulator supply.

 

However, as mentioned, floating the filaments is perfectly OK so this is not really necessary - it's just feeding my OCD .  If you float the filament windings they should automatically adjust to a safe limit.  Otherwise, if you want to bias the filaments, you can use the 590k and 133k resistor string from my shunt reg supply across the B+ to set the input filament voltage, and the 10k and 100k resistor string for C- to set the output filament voltage.
"

[mwl168]

Hi JimL:

I am building the SRX Plus and have been thinking about how to adjust the completed amplifier for optimal operating condition. I compiled the following procedure by reading and parsing through the information in your earlier posts on the SRX Revisited thread. Can you verify and advise if it’s correct?   

1. Set input stage current source to about 1.1 mA by adjusting the 500R pot and measure the voltage drop of the 100R resistor in series between the -20V input and the 500R pot. (Assuming 300K plate resistor for the upper 12AT7 as marked on the board. If you use 250K plate resistor then it should be 1.3mA)
2. Set the output current load for 7 mA by adjusting the 100R pot and measure the voltage drop of the 100R resistor in series between the 5.1K output resistor and the 100R pot.
3. Set the output current sink for about 17 mA by adjusting the 20R pot and measure the voltage drop of the 100R resistor in series between B- and the 20R pot.
4. Set output balance close to 0 VDC by adjusting the 5K pot.   Set the INPUT balance to 0 VDC by adjusting the 5k pot.  You are trying to get the upper 12AT7 plate voltages to be the same.  An easy way to do this is to connect your meter leads between the "bottom" of the two 12AT7 plate resistors and adjust the 5k pot to zero volts.  
5. Set output offset close to 0 VDC by adjusting the output current sink pot (20R). 

Also, below is a quote from your post on May 6, 2015. Can you advise where and how this fine-tuning fit in?

Thanks!

“…the published schematic has adjustments in the output current loads, this can in fact be used to vary the cathode-to-plate voltage individually for each output section by slightly altering the current running through each section.  The concept is, start with the output current loads fixed, adjust the cathode current sink to approximately zero the output plates, then vary the current loads to adjust the offset between + and - voltages for each channel.  The three adjustments are interactive, so, it would probably be best to adjust one of the current loads to partially decrease the offset, the adjust the other in the opposite direction to further decrease the offset until that is balanced, then adjust the current sink to zero both plates…”

 

EDIT: procedure corrected based on JimL's response in the next post where he also explains how to adjust for output balance if necessary.

Edited April 9, 2016 by mwl168