SRX-Plus Build Thread

[JimL]

The SPX431 is Exar's improved version of the TL431 adjustable IC shunt regulator.  It has about 10 dB less noise than the basic version and is available at Mouser.  The TL431 is the control element that sets the voltage of the regulator.  TIL731 is something completely different and will not work in this circuit.

Edited March 28, 2016 by JimL

[JimL]

Also remember with a shunt regulator that you MUST have the high voltage MOSFETs attached to the heatsink before testing it or it WILL BLOW UP!  I used four 20k 20 watt resistors in series parallel to make a 20k 80w dummy load when testing out the high voltage section of my regulator.

[jose]

Jim thanks for the recommendation. I will take precautions when starting the psu.

However, I do not understand one thing. On the scheme of your PSU I can see the SPX431 but on the layouts PCB Dr. Gilmore uses the TIL731. You say that it didn´t work in this circuit ... I´m not sure. Please, Is it correct?

Edited March 28, 2016 by jose
add PSU file

[kevin gilmore]

my bad on the mislabeling.  Hey... people are supposed to check this stuff.

updated board file in a minute.

[jose]

Sorry Dr. Gilmore . I guess I´m one of the first to make the psu ? 

Fortunately It's easy to fix.

 

[JimL]

Well, I missed that on the PS board also. 

[MLA]

As did I, and I thought I checked... 

Anyway, good that you caught it, and since you're very likely the first builder, please report if you run into any other issues...

[mwl168]

I have received requests to share my adjustment procedures so I thought I would post it here in case others are interested. These steps are based on combination of compiling the excellent information JimL has posted and my own experience.

All the trim pots except the 5K ones are for adjusting current setting of the current sources and sinks. And since they all use DN2540 which has quite a variance among samples, these setting will vary but the pre-set values in step 1 should provide a good starting point. You'll then need to fine-tune them a bit.

Here are the adjustment/fine-tuning steps of the SRX Plus in sequential order:

1. Initial setup before first power on: (assuming 7mA of output current per triode of the 6SN7GTA/B)

• Output current source (the 4 100 ohm trim pots close to the large heatsinks): adjust the trim pots to 55 ohm if you are using 180 ohm in-series fixed resistor (250 ohm on silkscreen). These trim pots set the output current of the 6SN7GTA/B
• Output current sink (the 2 20 ohm trim pots between the upper 12AT7 tubes): adjust the trim pots to 12.5 ohm if you are using 75 ohm in-series fixed resistor (102 ohm on silkscreen). These trim pots set the current sink of the 6SN7GTA/B.
• Input current sink (the 2 500 ohm trim pots): adjust the trim pots to 410 ohm if you are using 1.2K ohm in-series fixed resistor (1.4K ohm on silkscreen) . These trim pots set the plate voltage of the upper 12AT7 (the ones closer to the 6SN7GTA/B).
• Adjust the 2 5K trim pots to be in the mid-point. These trim pots are used to balance the voltage of the plates of each of the two triodes of the upper 12AT7 so they are equal. 
• The 5K pot and the 500 ohm pot will interact with each other a bit.

2. Insert all the tubes and power up the amp. If you have a variac, I recommend that you use the variac to gradually bring up the voltage for the first time especially if you are using NOS tubes that have not been powered on for a long time.

3. Check to see if all voltage (filament supplies, B+/B-, and the -20VDC) are all as expected and all the tubes lit up as expected. If everything appears normal, let the amp warm up for about 10 minutes before carry on to next steps.

4. The goal of this step is to set the output current of the 6SN7GTA/B to the desired value (7mA). Measure the voltage drop of the 100 ohm resistors of the output CCS (the 4 resistors between the large heatsinks and the .1uf/400V caps by the edge of the PCB). They should measure about 0.7VDC (0.7VDC/100R = 7mA). Adjust the 100 ohm trim pots to get the 0.7VDC voltage drop on these 100 ohm resistors.

5. The goal of this step is to set the output current sink to 17mA, 3mA higher than the output current. Measure the voltage drop of the 100 ohm resistors of the current sink (the 2 resistors between the 20 ohm trim pots and the single .1uf/400V cap between the two lower 12AT7 tubes). They should measure about 1.7VDC (1.7VDC/100R = 17mA). Adjust the 20 ohm trim pots to get the 1.7VDC voltage drop on the 100 ohm resistors.

6. The goal of this step is to set the two plates of each of the upper 12AT7 to be equal and about mid-way between B+ and ground. So if you are using B+ of 360VDC, the mid way is 180VDC.

First attach the DMM probes to each of the two .22uf/1000V coupling caps where they are connected to the plates of the upper 12AT7 and measure the DC voltage between them. (in my case. I purposely left the leads of the coupling caps sticking out a bit so I can clip the DMM probes on them. With the 5K trim pot set at its mid-point, the measured DC voltage between the two plates tells you how well matched are the two triodes of the 12AT7 tubes used. If they are well matched it should be within 5 VDC or so. If it's more than 20VDC you may have problem**.) Adjust the 5K trim pot so the measured voltage is close to 0VDC.

Next, move one of the DMM probes and clip it to the power ground (leave one probe still connected to one of the coupling caps) and adjust the 500 ohm trim pot so the measure voltage is about 180VDC (half of the B+). 

The adjustment of the 500 ohm and the 5K trim pots will interact with each other so you'll need to go back and forth a bit.

7. Let the amp warm up for another 15 - 20 minutes for the tubes to reach their stable operating state then repeat the adjustments again. Measure the balance via the headphone jack like you do with other Stax amps. If the two triodes in each of the 6SN7GTA/B are well matched you should see around 1 or 2 VDC or so. And this is perfectly fine.

8. Measure the offset via the headphone jack like you do with other Stax amps. Adjust the 20 ohm trim pot (output current sink) to bring the offset as close to 0 VDC as possible.

9. Assuming all adjustments are carried out without problem, you are done and ready to listen to music now ? This is an all tube amp without servo so the offset and balance etc. will never be spot on and will vary and drift a bit every time you power on the amp. That’s to be expected.

10. There is an optional step which I have not done myself - you can adjust the 2 100 ohm output CCS trim pots to bring the balance of each channel to 0 VDC. I have not felt the need to do it (maybe I got lucky with the tubes I have). If you choose to do so, try to adjust both trim pots in a converging fashion instead of adjusting only one of the two pots. You'll also need to "coordinate" this with adjusting the 20 ohm pot in step 8. 

Hope this helps. JimL and others please chime in and correct me.   

  

** In my experience, with the 5K trim pot centered, if the voltage difference between the two plates of the 12AT7 is greater than 15VDC or so, you should still be able to adjust it to 0 VDC but you may encounter problem with buzzing noise. You can move the 12AT7 tubes around to minimize this voltage difference. In my experience, it's more important to have the lower 12AT7 (the one closer to the 5K trim pot) well matched to bring this voltage difference down.   

Edited January 19, 2020 by mwl168

[jose]

wooo, thanks mwl.

Now I have my SRX on "stand by". I'm working on other amps and I haven´t time for three amp at the same time.

[JimL]

So, the one thing I would do differently is to set the current sinks and loads to close to their desired value before plugging everything in.  The nice thing about the current sources and loads is that, due to their very high impedance, you can get close to the final values without using high voltage.  You can do the preliminary adjustment as soon as the amp board is built without hooking up the high voltage OR filament voltage supplies.

 

What you need is a voltage source of around 15-25 volts, so for example, 2 nine volt batteries in series connected with clipped wires will do.  You connect the positive lead to the "upper" end of the current source and the negative end to "lower" end.  

 

Specifically, for the output current loads:

Connect the positive end to the B+ terminal,

 connect the negative end to the junction between the 5.1 k output resistor and the 100 ohm test resistor

Measure the voltage across the test resistor and adjust the trim pot to get 0.70 volts

 

For the output current sink:

Connect the positive lead to the junction between the 220k resistor string and the  10M90s MOSFET 

Connect the negative lead to the junction between the 500k grid resistors and the 100 ohm test resistor.

Measure the voltage across the test resistor and adjust the trim pot to get 1.70 volts.

 

Fot the input current sink:

Connect the positive lead to the junction between the 5k trim pot and the "top" DN2540.

Connect the negative lead to the -20v terminal

Measure the voltage across the test resistor and adjust the trim pot to get 0.11 volts  (if you are using 300k plate resistors for the 12AT7) or 0.13 volts (if you are using 300k plate resistors for the 12AT7).

 

This will get you close to the final values.  If you can't get to these values then you will need to adjust the value of the fixed resistor which is used with the trimpot to adjust the current.  If the measured voltage across the test resistor is too high, you will need to increase the value of the fixed resistor, and if the measured voltage is too low, you will need to decrease the value of the fixed resistor, to shift the current range so that the trim pot can be used to achieve the desired current.

[mwl168]

Thanks Jim. 

What you suggested is exactly what I did before I powered up my SRX Plus for the first time. It worked like a charm (thanks SorenB).

The current will drift upwards slightly as the amp warms up and because of the variance of the parts the trim pots will likely end up at different values. The preset values I suggested in step 1 are based on actual measurement I took after the amp was stable. The good news is that once these CCS are set, they remain very stable.

Edited April 30, 2016 by mwl168

[JimL]

Jose, aka Strawhat, noticed another error on the shunt power supply board.  There is a missing connection between the input ground terminal and the output terminal ground terminal.  The easiest way to fix this is to run a wire from the center tap of the HV transformer to both the input ground terminal and the output ground terminal.  Thanks to Jose for catching this.

[kevin gilmore]

so i'm not sure that is what you want because then the minus supply is completely unregulated which is the problem when you try to split the ground with only one regulated power supply. in effect it shorts out the 2 x 365k resistors.

http://gilmore.chem.northwestern.edu/srxshunt3.zip

has a position to short the unreg ground and regulated ground.

[JimL]

Well, maybe I'm confused, but the regulator is regulating the voltage between the B+ and B- rails but is not directly regulating either B+ or B-, other than the fact that the resistor chain across the output caps is dividing the voltage between B+and B- with respect to ground.  So, the total voltage between B+ and B- are regulated but they "float" around ground with the average value set by the dividing resistor string.  This is as opposed to KGs designs (as well as many others) where both B+ and B- are "hard" regulated with respect to ground.  This is a disadvantage of this design topology.  Does that make any sense?

Edited May 1, 2016 by JimL
clarification

[jose]

Resistors, please check.

100 O, 0.5 W gate stoppers for DN2540, and sense resistors		18		660-MF1/2CCT52R1000F
75 O, 0.5 W (source resistor for output cathode sink)		2		660-MF1/2CC75R0F
180 O, 0.5 W (source resistor for output current load)		4		660-MF1/2CC1820F	*182 R
500 O. 0.5 W		4		660-MF1/2CC4990F	*499R
1.0 kO, 0.5 W gate stopper for 10M90S		6		660-MF1/2CC1001F
1.2 kO, 0.5 W (source resistor for input current sink)		2		660-MF1/2CC1211F	*1,21K
2.0 kO, 0.5 W		4		660-MF1/2DCT52R2001F
5.1 kO, 1 W/500 V		4		294-5.1K-RC
124 kO, 0.5 W, 350 V rated		8		660-MF1/2CC1211F
220 kO, 2 W, 500 V rated		8		660-PCF2C224K
300 kO, 0.5 W, 350 V rated		4		660-MF1/2CC3013F	*301K
500 kO, 0.5 W, 350 V rated		4		660-MF1/2CC4993F	*499K

And trimpot...  Im´not sure about last reference (5K) please, any alternative?

20 O multiturn trimmer for output current sink—set current to 17 mA		2		652-3299W-1-200LF
100 O multiturn trimmer for output current load—set current to 7 mA		4		2-3299W-1-101LF
500 O multiturn trimmer for input stage current sink—set current to 1.33 mA		2		652-3299W-1-501LF
5 kO, 0.75 W multi-turn trimmer for input stage balance		2		652-3252W-1-502LF

[mwl168]

I am fairly certain I used Bourns 3296W-1-502F for the 5K trim pots. The rest of the trim pots I use are all 3296W series. At a glance, I cannot make out major difference between 3299 and 3296 series. Both are rated at 1/2W. 3252 series appears to be larger in size and quite a bit more expensive.

[jose]

Thanks mwl.

[JimL]

A couple weeks ago one of the 12AT7s in my SRX Plus failed, and in troubleshooting it eventually took out both input sections.  On analyzing the failure it appears that I may have been just a bit too much of a cheapskate.  Therefore I have slightly revised the circuit so that at least a failure won’t take out both channels, and hopefully won’t even take out one.

 

The revision is pretty straightforward.  Rather than supplying both input current source tails with a single resistor from B-, I substituted a 220k/500V rated resistor (270k/500V if you are using the original 300k plate resistors) from the B- rail to the input tail current source of each channel (which was previously marked on the schematic as being connected to -20v).

 

The revised schematic is shown (ignore the +/-340 volt numbers, they should be +/-325 v).  This resistor takes the place of the 10k and100k resistor chain in the shunt power supply that ran from B- to -C.  This means that C- in my original power supply is no longer connected to anything, and the only connections between the high voltage power supply and amp boards are B+, B- and ground.

 

For those using my shunt regulated current supply, you have to make a couple changes on the amp board.  First, you have to cut the connection between the -20 volt terminal and one of the 100 ohm sense resistors, and then you have to cut the connection between that 100 ohm resistor and the other 100 ohm sense resistor.  Then you have to connect each of those two sense resistors to a 220k resistor, which then runs to the B- rail – probably the closest place to do this on the amp board is the connecting point for the 100 ohm sense resistors on the output tail current sources – a bit of a kludge but it will work.

 

In addition, I substituted a 10M90S in place of the upper DN2540 on the input current source. This modestly improves its performance, but more importantly should prevent the input current source from failing due to overvoltage.

 

This modification does not need to be done for those of you who built the SRX Plus circuit using a Gilmore regulated PS, as that supplies the -15 volts that can be used for the input current sources.

[kevin gilmore]

srx6rev2 board file posted with the latest fixes, someone please check

 

[Arthrimus]

For those of us who will be building the amp on a KGBH power supply, what specs are we looking for as far as the power transformer? Voltage, current, etc? Thanks.

[JimL]

So, the amp consumes around 60-65 watts, similar to a KGSS.  A 100 VA transformer should probably be adequate, although higher won't hurt, with the output voltage running around +/-375 volts raw voltage to have some margin during the summer when the AC voltage may drop 10%.

[Arthrimus]

So something like this Antek http://www.antekinc.com/as-2t350-200va-350v-transformer/ wouldn't give me enough headroom. I guess I'd have to step up to one of the 400VA models to get anything higher than 350V. 

Just to clarify, can center tapped transformers be used with the KGBH power supply or do they have to have separate high voltage secondaries? 

Also I was curious what the max transformer voltage would be safe to use. Antek has some 430V transformers in stock right now. would that be suitable for this application or would that be too much excess for the power supply? Thanks.

Edited June 14, 2016 by Arthrimus

[JimL]

Um, depends on whether you are using cap input or choke input filter.  I was referring to the filtered voltage out, not the AC voltage out.  If you are using cap filter then something around +/-270 volts AC RMS should work, if you use a choke filter then yeah, you need 400 volts AC RMS.  I believe the KGBH PS has the rectifiers on the board and it is a cap filter board so something around +/-270 volts AC RMS should do it.  My design uses a choke filter (although KG's board design has the rectifiers on board) so if you wanted to have choke filter input you should use around +/-400 V RMS.

Edited June 14, 2016 by JimL

[Arthrimus]

Thanks so much for the help. I intend to use the KGBH board in it's standard cap filtered configuration so that's the info I needed.

[kevin gilmore]

you cannot use a center tapped transformer with the kgbh power supply, you need 2 seperate high voltage windings