Best method is to measure the voltage drop across R6 so you can calculate the current and dissipation.
IT'S ALIIIIIIVE! (and so am I)
Thanks so much to everyone who has been generous with their time and advice in this thread. You guys are great, and I'm looking forward to being a contributor once this build is finalized. As we know, wisdom comes from folly and experience, both of which were dealt in spades tonight as the result of some non-trivial, self-caused issues.
I finished the checkout procedure and did some listening for about 90 minutes tonight, about 15 minutes on the el cheapo RadioShack speakers I got for testing, and then about 75 minutes on the good speakers.
Here's what happened:
* The amp sounds very nice. I am pleased and planning improvements already.
* I biased the 300Bs to 74mA, which brought the B+ down to 372v.
* The 5842s' plate voltage tested at 163v yesterday.
* There is a constant hum (I believe 60Hz) that doesn't change volume, and which I can hear from the listening position when music is not playing
* The volume is adequate for normal listening but I'd like a little more (the speakers are 91.5db MTMs). The dial was set to a couple of notches below maximum volume and at maximum volume for most of the session. Perhaps I should switch the 50k attenuator that's in there with a 25k or 10k model.
* R6 is running hot (166-177*F, sometimes up to 226*F)
* U1 (voltage regulator) also runs hot (~175*F), but it never shut the amp down in 90 minutes of listening
* I blew the GZ34 regulator tube when I turned the amp off (after 90 minutes listening) and then back on about 20 seconds later. Is that normal?
* I connected everything back up to my solid state amp and there is now no sound coming out of the right speaker, which I believe to have happened after the GZ34 blew
** Lessons learned: Listen to your crappy test speakers for several hours and for two or three sessions before switching to your good ones; learn the amp's behavior and don't try to do too much at once.
*** Forehead smacking moment: I just looked at the parts list and read the note about using a 36k 3W resistor at R14 and R25 for 350-400V applications. When I soldered the board I inserted the 20K one in the parts list and had been planning to change it but forgot to. Maybe that's why R6 is running so hot?
Since my OPTs are 5K:8ohm:15w, I was aiming for a B+ of 390-400v with a bias of 80-90mA on the output tubes (Psvane 300B hi-fi series). One reason for the lower-than-target voltage/bias could be that I'm using the 150 ohm/5W resistor at R4 instead of a choke. I did this because once the amp is burned in I want to replace the resistor with a choke so I can hear the difference. After adding the choke I'm planning to add a motor run cap at C5 so I can hear that difference too.
After 90 minutes of listening, I was curious to know what the plate voltage on the 5842s was doing, so I turned off the amp, switched the meters to those tubes, and turned it back on. After a couple seconds there was a spark and "fzzt" in the GZ34 regulator tube and the voltage in the 5842s was at 10v, so I shut it off. Not sure what happened there, and I hope it didn't cause damage to other parts of the amp. I will re-test the circuit tomorrow before doing anything else.
It was very nice while it lasted, but I should have let it cool for the night and come back to it tomorrow. Now I have to get my speaker repaired (I'm guessing something blew in the crossover) or attempt it myself.
Thanks so much to everyone who has been generous with their time and advice in this thread. You guys are great, and I'm looking forward to being a contributor once this build is finalized. As we know, wisdom comes from folly and experience, both of which were dealt in spades tonight as the result of some non-trivial, self-caused issues.
I finished the checkout procedure and did some listening for about 90 minutes tonight, about 15 minutes on the el cheapo RadioShack speakers I got for testing, and then about 75 minutes on the good speakers.
Here's what happened:
* The amp sounds very nice. I am pleased and planning improvements already.
* I biased the 300Bs to 74mA, which brought the B+ down to 372v.
* The 5842s' plate voltage tested at 163v yesterday.
* There is a constant hum (I believe 60Hz) that doesn't change volume, and which I can hear from the listening position when music is not playing
* The volume is adequate for normal listening but I'd like a little more (the speakers are 91.5db MTMs). The dial was set to a couple of notches below maximum volume and at maximum volume for most of the session. Perhaps I should switch the 50k attenuator that's in there with a 25k or 10k model.
* R6 is running hot (166-177*F, sometimes up to 226*F)
* U1 (voltage regulator) also runs hot (~175*F), but it never shut the amp down in 90 minutes of listening
* I blew the GZ34 regulator tube when I turned the amp off (after 90 minutes listening) and then back on about 20 seconds later. Is that normal?
* I connected everything back up to my solid state amp and there is now no sound coming out of the right speaker, which I believe to have happened after the GZ34 blew
** Lessons learned: Listen to your crappy test speakers for several hours and for two or three sessions before switching to your good ones; learn the amp's behavior and don't try to do too much at once.
*** Forehead smacking moment: I just looked at the parts list and read the note about using a 36k 3W resistor at R14 and R25 for 350-400V applications. When I soldered the board I inserted the 20K one in the parts list and had been planning to change it but forgot to. Maybe that's why R6 is running so hot?
Since my OPTs are 5K:8ohm:15w, I was aiming for a B+ of 390-400v with a bias of 80-90mA on the output tubes (Psvane 300B hi-fi series). One reason for the lower-than-target voltage/bias could be that I'm using the 150 ohm/5W resistor at R4 instead of a choke. I did this because once the amp is burned in I want to replace the resistor with a choke so I can hear the difference. After adding the choke I'm planning to add a motor run cap at C5 so I can hear that difference too.
After 90 minutes of listening, I was curious to know what the plate voltage on the 5842s was doing, so I turned off the amp, switched the meters to those tubes, and turned it back on. After a couple seconds there was a spark and "fzzt" in the GZ34 regulator tube and the voltage in the 5842s was at 10v, so I shut it off. Not sure what happened there, and I hope it didn't cause damage to other parts of the amp. I will re-test the circuit tomorrow before doing anything else.
It was very nice while it lasted, but I should have let it cool for the night and come back to it tomorrow. Now I have to get my speaker repaired (I'm guessing something blew in the crossover) or attempt it myself.
Last edited:
Yeah, it's normal if you only wait 20 seconds to power back up. 3-5 minutes min. would have probably saved your rectum-frier (and anything else that may have died)... A couple of cheap dmm's would have helped too, no need to power off. Installing a choke could take care of the hum.
^ What he said! I have used cheap JJ rectifiers in the past and they don't tolerate that. Always keep a spare on hand. I was lucky to score a few NOS Mullards and they are much more tolerant and last longer BUT when you pay that much for a rectifier, you are also more careful! 😛
Glad to hear it is/was up and running!
As for speakers, I picked up a set of 100W 8 ohm resistors on eBay that I mounted to a piece of plywood with banana plugs to use as dummy loads for the first bring-up and make sure everything is idling nicely while checking voltages BEFORE plugging into my speakers. Always double-check DC offset at the speaker terminals too and make sure it is zero (or millivolts) before plugging speakers in.
Last edited:
Thanks, guys. This is frustrating, depressing, and exciting all at the same time.
This week I'll test down the schematic chain again to make sure nothing else is blown, and restart the check out process from scratch once I've got a new rectifier.
I've been shopping around for rectifier tubes online and am noticing new production Mullards (made in Russia) that are supposed to come up slowly and extend the life of the other tubes in the system. They're around $38 or so plus shipping. Any experience with these? Any other recommended brands?
This week I'll test down the schematic chain again to make sure nothing else is blown, and restart the check out process from scratch once I've got a new rectifier.
I've been shopping around for rectifier tubes online and am noticing new production Mullards (made in Russia) that are supposed to come up slowly and extend the life of the other tubes in the system. They're around $38 or so plus shipping. Any experience with these? Any other recommended brands?
I've used Sovteks several times... they seem to be fairly reliable and it's what George has used. I also install ICL's on all projects... slows the warm-up even more.
They should be fine. New production Mullard, Sovtek, EH, Tung Sol are all New Sensor and are decent tubes. JJ is hit or miss and stay away from unbranded Chinese tubes. The best of the best for a bulletproof rectifier is an old Mullard. They will run between $50 and $100 each on eBay if you take your time and shop around. Sometimes it is a better deal to buy a pair.
Sovtek's are cheap and reported to be very reliable. I have had one for five plus years with no issues. I think I paid around fifteen dollars for it.
Edit: I think all indirectly heated rectifiers come up slowly enough to prolong tube life. I built my SSE with a standby switch to use it with 5U4's, but just use the recommended 5AR4 without the standby switch because the warm-up is much gentler on the rest of the tubes.
Edit: I think all indirectly heated rectifiers come up slowly enough to prolong tube life. I built my SSE with a standby switch to use it with 5U4's, but just use the recommended 5AR4 without the standby switch because the warm-up is much gentler on the rest of the tubes.
Last edited:
Thanks again. While the new GZ34 is on its way, I'd like to ask the forum for advice on how I can get the B+ up closer to 390-400v.
George's instructions recommend using 36K resistors at R14 and R25 for 350-400v applications. More resistance there would decrease the B+, though, wouldn't it? Should I keep the 20K resistors I've got in there now?
Also, do you think adding a choke in place of R4 and/or a motor run cap in parallel with C5 would boost B+ as well?
I got 5K:8ohm OPTs for the purpose of using higher B+ and output tube current.
Before I attach this amp to speakers again, I'm going to look at it through an oscilloscope with a local tube amp engineer who's offered to help out. I hope to share some graphs and data at the end of the week.
George's instructions recommend using 36K resistors at R14 and R25 for 350-400v applications. More resistance there would decrease the B+, though, wouldn't it? Should I keep the 20K resistors I've got in there now?
Also, do you think adding a choke in place of R4 and/or a motor run cap in parallel with C5 would boost B+ as well?
I got 5K:8ohm OPTs for the purpose of using higher B+ and output tube current.
Before I attach this amp to speakers again, I'm going to look at it through an oscilloscope with a local tube amp engineer who's offered to help out. I hope to share some graphs and data at the end of the week.
Last edited:
You can increase C4 but be careful not to exceed the input capacitance of the GZ34 (60uF). You can also reduce R4 or the equivalent choke resistance. I suggest downloading PSU Designer II from Duncan Amps and running some simulations to see if you can achieve your target voltage with the transformer you have. Just be cautious of the caps on the B+ supply. the rule of thumb is don't exceed 80% of the rated voltage.
OK, thanks craigtone. I'll download PSU Designer and start playing with it.
I'm thinking about the excessive heat coming from R6. If I read the schematic correctly, the path goes along the bias supply from R6 (which in my amp gets super hot) to C7, R7, and then R14/R25 into the MOSFETs. Am I right in thinking that swapping out the 20K 3W resistors for the 36K ones at R14/R25 would cause them to take on more load, reducing the load (and therefore the temperature) of R6?
Perhaps related to that, JP expressed concern about whether I'm using the right MOSFETs here. Since R14/R25 feed the MOSFETs, perhaps if I put the 36k resistors in there the MOSFETs (Toshiba 2SK3565QM) will be happier?
I'm thinking about the excessive heat coming from R6. If I read the schematic correctly, the path goes along the bias supply from R6 (which in my amp gets super hot) to C7, R7, and then R14/R25 into the MOSFETs. Am I right in thinking that swapping out the 20K 3W resistors for the 36K ones at R14/R25 would cause them to take on more load, reducing the load (and therefore the temperature) of R6?
Perhaps related to that, JP expressed concern about whether I'm using the right MOSFETs here. Since R14/R25 feed the MOSFETs, perhaps if I put the 36k resistors in there the MOSFETs (Toshiba 2SK3565QM) will be happier?
IIRC, going to 36K for R14,25 was to reduce the B- voltage on the MOSFETs, as the B- supply is taken from the same secondary as the B+ supply. IOW, if you use a high-voltage secondary for more B+, you're also increasing the B- voltage which is 1) not needed and 2) harder on the MOSFETs.
For R6, again, measure the voltage drop across it and figure out what its dissipating. Heat will depend on dissipation and surface area. This means hot isn't necessarily bad.
My MOSFET concerns was more about where you got them. For some reason a lot of folks get them off eBay, of which most are fake.
How is your PT primary wired? Black and brown?
For R6, again, measure the voltage drop across it and figure out what its dissipating. Heat will depend on dissipation and surface area. This means hot isn't necessarily bad.
My MOSFET concerns was more about where you got them. For some reason a lot of folks get them off eBay, of which most are fake.
How is your PT primary wired? Black and brown?
I got the MOSFETs from Digikey and picked them based on their low and flat Crss rating, a characteristic that George advocated here.
PT primary is black and brown.
I'll replace the 20k resistors with 36k at R14/R25 to reduce B- and go a little easier on the MOSFETs.
PT primary is black and brown.
I'll replace the 20k resistors with 36k at R14/R25 to reduce B- and go a little easier on the MOSFETs.
I've been working with a local tube amp engineer to troubleshoot the excessive heat coming from some of the resistors and other sources. When measuring voltages around the amp, the B- was close to -300V, but the schematic indicates it should be -150V.
According to boywonder, in response to acceler8's similar B- measurements, more than -300V is normal:
http://www.diyaudio.com/forums/tubelab/207272-tubelab-se-questions.html#post3088133
I'm just wondering if other 300B builders have found the same. It seems if you could get the B- down to -150V, then the 300Bs and the MOSFETs would be happier.
In order to bring the B- closer to -150V, the engineer had me remove one of the hexFREDs at D3 and changed R7 from a 100k to a 10k resistor. It is now about -170V.
Tonight I started the checkout procedure from the beginning. Everything looked good until the driver tube plate voltage stage, when I could get the left channel to 165V but the right only to 156V. I did that test a second time and the results were the same. They are supposed to be set to 175V. The highest I had in the original configuration last week was 165V each.
Because of the 9V difference between channels, I decided not to proceed to the output tube bias step, and have left the 300Bs out of the amp
Here is the series of test results, in order of completion:
No tubes
* B+ = -47
* B- = -167
* Filament voltage (pins 1 & 4) = 4.94
* Filament current (pins 1 & 4) = 4.96
With rectifier
* B+ = 468
* B- = -161
Output tube grid voltage (with rectifier, without 300Bs)
* B+ = 472
* B- = -168
* L grid (pin 3 to ground) = -565
* R grid (pin 3 to ground) = -567
With 5842s
* B+ = 457
* B- = -179
* L plate (C11 to ground; tip jacks) = 165
* R plate (C9 to ground; tip jacks) = 156
R channel plate voltage is 9V off, and the output grid voltages seem extreme (but maybe not? It is after all the most negative voltage I could get with the trim pots as per the checkout instructions.)
So, I'm wondering if I should put that hexFRED back in, and if I should do something to raise the plate voltage of the 5842s, like change the value of R3?
Thanks for any advice you guys might have.
PS -- JP: the voltage drop across R6 is 241V with the GZ34 and 5842s but without the 300Bs.
According to boywonder, in response to acceler8's similar B- measurements, more than -300V is normal:
http://www.diyaudio.com/forums/tubelab/207272-tubelab-se-questions.html#post3088133
I'm just wondering if other 300B builders have found the same. It seems if you could get the B- down to -150V, then the 300Bs and the MOSFETs would be happier.
In order to bring the B- closer to -150V, the engineer had me remove one of the hexFREDs at D3 and changed R7 from a 100k to a 10k resistor. It is now about -170V.
Tonight I started the checkout procedure from the beginning. Everything looked good until the driver tube plate voltage stage, when I could get the left channel to 165V but the right only to 156V. I did that test a second time and the results were the same. They are supposed to be set to 175V. The highest I had in the original configuration last week was 165V each.
Because of the 9V difference between channels, I decided not to proceed to the output tube bias step, and have left the 300Bs out of the amp
Here is the series of test results, in order of completion:
No tubes
* B+ = -47
* B- = -167
* Filament voltage (pins 1 & 4) = 4.94
* Filament current (pins 1 & 4) = 4.96
With rectifier
* B+ = 468
* B- = -161
Output tube grid voltage (with rectifier, without 300Bs)
* B+ = 472
* B- = -168
* L grid (pin 3 to ground) = -565
* R grid (pin 3 to ground) = -567
With 5842s
* B+ = 457
* B- = -179
* L plate (C11 to ground; tip jacks) = 165
* R plate (C9 to ground; tip jacks) = 156
R channel plate voltage is 9V off, and the output grid voltages seem extreme (but maybe not? It is after all the most negative voltage I could get with the trim pots as per the checkout instructions.)
So, I'm wondering if I should put that hexFRED back in, and if I should do something to raise the plate voltage of the 5842s, like change the value of R3?
Thanks for any advice you guys might have.
PS -- JP: the voltage drop across R6 is 241V with the GZ34 and 5842s but without the 300Bs.
Last edited:
Re: The 5842's - I would set them to about 150 volts and call it good. IIRC, I had one that wouldn't get above 145. I don't think it's that critical.
Or you can adjust their cathode resistance.
Win W5JAG
Or you can adjust their cathode resistance.
Win W5JAG
OK, thanks. I put in the 300Bs and re-tested (btw, w/8 ohm dummy loads on the output binding posts and shorting plugs on the input RCA jacks).
The 5842s' plate voltage increased but there is still a 10v difference between them. How to correct that?
Bias is now higher at about 79mA per channel but B+ is still ~370. Was hoping to get up to B+ of 390 with bias between 80-90mA. Maybe a choke with lower resistance at R4 (current one is 150 ohm, 6H, 200mA) would help raise the B+?
Also, some of the resistors and ICs are still getting really hot. R6, R7, the voltage regulator and the MOSFETs were all at 140*F after the amp ran for a few minutes.
Here are the new measurements:
* B+ = 370
* B- = -170
* L bias (300B) = 79.3
* R bias (300B) = 79.6
* L plate (5842) = 169
* R plate (5842) = 159
* C4 voltage (positive lead to ground) = 399
* C5 voltage (positive lead to ground) = 367
* C6 voltage (negative lead to ground) = -414
* C7 voltage (negative lead to ground) = -170
* D2 voltage (close lead of R5 to ground) = -418
* Voltage drop across R6 = 241
* Voltage drop across R7 = -170
The 5842s' plate voltage increased but there is still a 10v difference between them. How to correct that?
Bias is now higher at about 79mA per channel but B+ is still ~370. Was hoping to get up to B+ of 390 with bias between 80-90mA. Maybe a choke with lower resistance at R4 (current one is 150 ohm, 6H, 200mA) would help raise the B+?
Also, some of the resistors and ICs are still getting really hot. R6, R7, the voltage regulator and the MOSFETs were all at 140*F after the amp ran for a few minutes.
Here are the new measurements:
* B+ = 370
* B- = -170
* L bias (300B) = 79.3
* R bias (300B) = 79.6
* L plate (5842) = 169
* R plate (5842) = 159
* C4 voltage (positive lead to ground) = 399
* C5 voltage (positive lead to ground) = 367
* C6 voltage (negative lead to ground) = -414
* C7 voltage (negative lead to ground) = -170
* D2 voltage (close lead of R5 to ground) = -418
* Voltage drop across R6 = 241
* Voltage drop across R7 = -170
Last edited:
I would set the plates of the 5842 equal at the limit of what you can get . If you want to get them higher it seems you would go after the 510ohm resistor paralleled to the trim pot r10/21
My B- was a lot more negative then George recommended and I adjusted some values to bring it into line...I found I liked the amp with the B- more negative.
r6 10k 241volts is dissipating almost 6 watts. If you up this value you will lower dissipation and make B- less negative.
r7 100k 170 volts is dissipating .29watt
I added bigger heatsinks to the semiconductors to be safe. I don't know that this was necessary but it made me feel better.
My B- was a lot more negative then George recommended and I adjusted some values to bring it into line...I found I liked the amp with the B- more negative.
r6 10k 241volts is dissipating almost 6 watts. If you up this value you will lower dissipation and make B- less negative.
r7 100k 170 volts is dissipating .29watt
I added bigger heatsinks to the semiconductors to be safe. I don't know that this was necessary but it made me feel better.
The 5842's aren't identical. Don't worry about it.
Set them to a common value in their range and be done with it.
Win W5JAG
edit: I burned up one of the resistors in the negative line on mine, and I replaced it with a higher wattage part. Don't recall which one it was, but if you look back at the pic of my board, you can see that one of the resistors was replaced - you can see the rework marks on the board.
second edit: you may be able to tweak your B+ up in a couple of different ways: first, you can adjust the value of the first capacitor upward in the CLC filter to see if that raises the voltage where you want it. Second, you can ditch the tube and go with a solid state tube replacement, which will increase the voltage proabably about 10%. You could then adjust it back down by decreasing the value of the first capacitor in the CLC filter.
There wil be an interaction between the current draw and power supply output voltage - as you draw more current, the voltage will be drawn back down, and vice versa. There are rectifer tubes that have a relatively constant voltage output as current increases or decreases, but they have four pins and contain hazardous substances. But they do exist.
I wanted a lower B+ in my TSE than I could achive with my PT, so I omitted the first capacitor entirely and used an LC filter onfiguration on my TSE.
Set them to a common value in their range and be done with it.
Win W5JAG
edit: I burned up one of the resistors in the negative line on mine, and I replaced it with a higher wattage part. Don't recall which one it was, but if you look back at the pic of my board, you can see that one of the resistors was replaced - you can see the rework marks on the board.
second edit: you may be able to tweak your B+ up in a couple of different ways: first, you can adjust the value of the first capacitor upward in the CLC filter to see if that raises the voltage where you want it. Second, you can ditch the tube and go with a solid state tube replacement, which will increase the voltage proabably about 10%. You could then adjust it back down by decreasing the value of the first capacitor in the CLC filter.
There wil be an interaction between the current draw and power supply output voltage - as you draw more current, the voltage will be drawn back down, and vice versa. There are rectifer tubes that have a relatively constant voltage output as current increases or decreases, but they have four pins and contain hazardous substances. But they do exist.
I wanted a lower B+ in my TSE than I could achive with my PT, so I omitted the first capacitor entirely and used an LC filter onfiguration on my TSE.
Last edited:
These suggestions make sense. However, the tubes used to have equal voltages of 163V with both pots set to maximum, and now they don't, which is concerning to me.
Also, this datasheet for the 5842 indicates a plate voltage range of 130-150V, much less than the 175V spec'd in the TSE design. Shouldn't the plate voltage be set below 150V?
I'm planning to swap in a 7k 10w resistor at R7 tonight, to see how it affects B- and other measured components. If it looks good then I might swap in 25w chassis mount resistors for those positions, as per the suggestion of the engineer who's been helping me.
He also recommends swapping in this STMicroelectronics STW20N65M5 MOSFET for its higher Vgs (gate-source voltage) and power dissipation ratings, which if I understand correctly will block more incoming voltage and allow it to run cooler. I notice that it has a lower, flatter Crss than the Toshiba 2SK3565 I've been using, too.