You can analyze this yourselves. There is no high-value capacitor at the output, and the ESR of the existing one is increased to maintain the other task (damping). The current supply during the startup (there is NO instant input voltage at the start up in the real life so your initial assumption is not correct) is enough to maintain rather low voltage across the MOSFET taking into consideration the gradually increasing voltage across the capacitor due to the charging current. It can be easily modelled, but I measured this with the real SSHV v.1.
I asked him. He didnt react. But wrote SSHV before, not SSHV2.
The link is live, uploaded by Salas and provided by DiyAudio. If realised that it's wrong, then should have been deleted and readers been warned. I didn't find any sign suggesting that it is obsolote, Salas insisted it is OK on the first pages. Nobody can be expected to read through 500 pages.
Yes, after the settling, if can be done. But in the first moment of turning on almost all input voltage drops on Q1. This is enough to kill it, if rises too fast. This because I suggested slowing down.
In this case 100 %, since it has actually happened to Roland.
Responded to this:
"I think we are discussing power supplies here. I do not know any "new" technology invented by JLH.
I also talk about stabilizer which is an amplifier for DC. JLH amp I refer to is not new, but relevant, if somebody wants to design a good, simple, efficient shunt regulator.
Post #1 has links to all versions. SSHV1 is adopted extensively in Oliver's Tubizators, maybe hundreds of builds, not sure about the number. Works for them. (Not the one with the BJTs in Vref). SSHV2 was my only recommended standalone pcb release. Thus Roland has SSHV2 if talking pcb which uses a 400V DMOS cascode input CCS. We don't have data why its blown, making assumptions, there are hundreds working fine out there as well. Its not a new project.
Effectively there is a high value cap: C2, multiplied by Q2, Q3. The charger of C2 is a faulty CCS formed by Q4, Q5, but it's current is actually not a real constant, since 0.6 V drop is not enough on them. Basically Q5 is always saturated, and current is determined by R5+Trim and Vbe of Q2, so 6 mA maximum, 4 mA typical . dVC2/dt is 4 mA/1 uF=4 V/ms. If input voltage rises faster (and this happens easily), then Q2 opens Q3, and shunts the input current.
As already pointed by Salas - actual link is on the first message of the thread and it points to correct schematics. In SSHV v.1 the CCS which determines the output voltage (wrongly pointed by you as "charger of the capacitor") consists of one 2SK170.
You didn't point to a particular amp or shunt regulator to look at.
You forgot another path for charging the C2 - emitter-base junction of Q2.
Dear readers, Im sorry for continuing this discussion! I thought I've made my point clear, but I have to react this for last time although 80 % Offtopic:
Does 1 link to a "correct" schematic in an other thread automatically discard more faulty attachments?
1: I didnt wrote about that one. Ive already linked the sch i referred to, in this thread, not in an other.
2: please try to prove that current does not charge C1! Good luck! In turn on process it is definitely the charger. And this discussion is entirely about turn on process.
Trivially. Did you ask for? Then I can do it in a different thread, since here it would have been offtopic. Many people knows that concept, the rest can ask without trying to prove it was nonexistent. Google it! Many results, but basically all the same concept.
Do you really think I could forget that path, what can destroy the PSU? Try to figure out how many microamps can it provide for C1 without damaging shunt FET! Considering I wrote about Q2 3 times, do you really think I forgot its base, or possibly I calculated its current and found it insignificant in case of non-fault conditions?
Does 1 link to a "correct" schematic in an other thread automatically discard more faulty attachments?
1: I didnt wrote about that one. Ive already linked the sch i referred to, in this thread, not in an other.
2: please try to prove that current does not charge C1! Good luck! In turn on process it is definitely the charger. And this discussion is entirely about turn on process.
Trivially. Did you ask for? Then I can do it in a different thread, since here it would have been offtopic. Many people knows that concept, the rest can ask without trying to prove it was nonexistent. Google it! Many results, but basically all the same concept.
Do you really think I could forget that path, what can destroy the PSU? Try to figure out how many microamps can it provide for C1 without damaging shunt FET! Considering I wrote about Q2 3 times, do you really think I forgot its base, or possibly I calculated its current and found it insignificant in case of non-fault conditions?
The link to the correct schematic is in the first message of THIS thread, not some "other". The schematic you are referring to is preliminary. It's a forum, there are many people who can help to polish an idea and to clean mistakes. The discussions help other people to eliminate problematic solutions in their devices. It's no good to delete them.
While I did not write that the circuit is not a charger, its primary role is different, that what I mentioned in my message.
Insignificant as a big picture - agree with you. At the start up the Q2 is in saturation region until the voltage on the output does not exceed some point. For this short time the base current can reach 40-50mA easily and without any damage to the PSU. But in the real life it does not occur, because of "slow start" in CCS (charging C1) - that is the main danger, because for a long time (probably close to 0.5s) the Q1 is in non-conducting state (always mix English terms "open" and "close", in Russian it means just opposite in the relations to transistors) and has the whole input voltage across it.
Didn't catch the reference to "shunt FET" - how can it be damaged?
I agree that power supplies are mostly amps, but even that knowledge can not eliminate the fact that their purposes and working conditions differ greatly. Just one point of difference - the nature of the load.
And it seems the effectiveness of the JLHs amps I found in the net is not greater (for class A amps) that everyone else's. Boiling in the audio side for a while I see that effectiveness is far from the main goals of people.
P.S. Unfortunately the discussion does not help Roland to mend his device, because he uses different schematic.
Last edited:
remove the shunt half of the regulator.Hi Salas,
My shunt reg is again working.
But, I used variac to slowly increase the voltage VLoad to 350Vdc.
I want to remove the variac now and connect it to Impasse.
Last time I did this - at power ON, Q1 popped and SSHV failed
Vin (unreg) is 380Vdc, is this too much to cause SSHV to fail, I don't think so?
If I remove the variac at switch ON, I'm afraid I may get the same problem
Do I need soft start circuit between Vin unreg and SSHV or any low value high power resistor in between? The reason I get rid of dropping resistor in the first place is to get rid of unwanted heat.
Can power ON at 380Vdc cause severe start up loading?
Can capacitor charging at Vunreg cause reg to fail therefore should I decrease capacitor value?
Please help.
regards,
Roland
Retain the CCS part of the regulator.
Attach a temporary load that is intended to draw the design current of the CCS. try 100r.
Apply a LOW DC voltage to the stripped down regulator.
Use your Variac into a low voltage transformer and gradually wind it up until the CCS is working properly and the load Vdrop indicates that the CCS current is actually passing through the load.
This can all be done at low and safe voltage. try using a 25Vac secondary The Variac will allow 0Vdc to 35Vdc to be applied.
prove the CCS works as intended.
Only after that testing is complete it worth taking the risk of applying hundreds of volts to the complete regulator.
But I would put in place a second stage of testing at low voltage.
Reduce the set point of the shunt regulator to around 30Vdc.
add this to the CCS. Now test with the Variac and the 25Vac transformer.
Prove the CCS and Shunt together are an effective low voltage regulator.
Now I would be more confident of feeding a hundred volts into the Salas regulator.
Hi Guys
Sorry for the (possibly very dumb question here) but I have a tubed output stage for DAC (a lampizator Amber balanced board) and was wanting to use two SSHV's to make a dual mono power supply configuration for each channel.
The Amber PCB layout is a single tube rectifier which then splits into two separate CRCRC banks - one to feed each channel on one board with a common ground.
My question: Can I use two SSHV's, each with their own tube rectifier-CRC-SSHV supply and the common ground? I am assuming here that both the SSHV's would have their 0vS lines connected to the common PCB ground but I suspect the circuits will be unstable - am I correct?
My understanding of the SSHV is basic.
Sorry for the (possibly very dumb question here) but I have a tubed output stage for DAC (a lampizator Amber balanced board) and was wanting to use two SSHV's to make a dual mono power supply configuration for each channel.
The Amber PCB layout is a single tube rectifier which then splits into two separate CRCRC banks - one to feed each channel on one board with a common ground.
My question: Can I use two SSHV's, each with their own tube rectifier-CRC-SSHV supply and the common ground? I am assuming here that both the SSHV's would have their 0vS lines connected to the common PCB ground but I suspect the circuits will be unstable - am I correct?
My understanding of the SSHV is basic.
Hello to everybody.
Need someone help to help me solve the issue I am facing.
SSHV2 kit on breadboard 3kohm load with CCS set to 50mA.
Raw DC 120V 30ma needed on load.
The trouble is that I can change the CCS set but I am not able to go over 77-78V on the output (when I turn R11 voltage does not change). I made some checks according to the guidelines I found on this thread:
CCS reading through test point has been confimed by ammeter.
CCS LV test: Vgs 1,1V with 7,85Vin (battery).
Q5 in circuit test (off power): between E-B I found 435ohm. Value changes when turning R11
Q4 in circuit test (off power): between E-B I found 1030 ohm.
Q6 in circuit check (off power): Rds=75ohm.
Red led works.
I have also tried with different CCS setting. 65mA, same load, Vout=77V Raw DC=90V
Any advice?
Need someone help to help me solve the issue I am facing.
SSHV2 kit on breadboard 3kohm load with CCS set to 50mA.
Raw DC 120V 30ma needed on load.
The trouble is that I can change the CCS set but I am not able to go over 77-78V on the output (when I turn R11 voltage does not change). I made some checks according to the guidelines I found on this thread:
CCS reading through test point has been confimed by ammeter.
CCS LV test: Vgs 1,1V with 7,85Vin (battery).
Q5 in circuit test (off power): between E-B I found 435ohm. Value changes when turning R11
Q4 in circuit test (off power): between E-B I found 1030 ohm.
Q6 in circuit check (off power): Rds=75ohm.
Red led works.
I have also tried with different CCS setting. 65mA, same load, Vout=77V Raw DC=90V
Any advice?
I'm working on a more compact version of the SSHV2 and had a question about the KSA1381. How can their heat dissipation be calculated and would it be suitable to replace them with a SOT223 package? If so, what about the PZTA92? It's higher hfe, but I'm not sure if that's a huge concern here. If not the PZTA92, any other recommendations?
I'd actually even prefer SOT89 if I can manage it, as it would match the DMOS devices.
My Vout will be 160V and 260V (so definitely < 300V), hence attempting the smaller parts.
I'd actually even prefer SOT89 if I can manage it, as it would match the DMOS devices.
My Vout will be 160V and 260V (so definitely < 300V), hence attempting the smaller parts.
Last edited:
VgsM2/R8=2.2mA*(Vo/2)=PdQ5 or PdQ4
PZTA92 seems likely to work. In what linearity or phase margin I don't know
Q4 Ohmic sounds low. Make sure about Q4, Q5 orientation, because plastic backed, spot their front by the lettering
The BF621 would be my preference then. Better? Same unknown factor? The high ft of the KSA1381 is hard to find.
Yeah, seem all about thermal performance, don't they. There is a SPICE model, not sure that helps:
http://www.nxp.com/documents/spice_model/BF621.prm
http://www.nxp.com/documents/spice_model/BF621.prm
Wow, thanks! Here is the PXTA92 which is a SOT89, probably similar performance?
http://www.nxp.com/documents/spice_model/PXTA92.prm
Many, many thanks!
http://www.nxp.com/documents/spice_model/PXTA92.prm
Many, many thanks!