• Disclaimer: This Vendor's Forum is a paid-for commercial area. Unlike the rest of diyAudio, the Vendor has complete control of what may or may not be posted in this forum. If you wish to discuss technical matters outside the bounds of what is permitted by the Vendor, please use the non-commercial areas of diyAudio to do so.

TSEII: all was going well until...

...until I smelled smoke.

This happened three days ago while I was measuring the V1/V2 voltages on pins 3 and 9. The amp is now exhibiting the same symptoms noted in this thread, except that this seemed to occur in both channels simultaneously, and I'm not measuring a short between the drain and source on either MOSFET. Also, R14 and 25 measure their nominal 20K and don't show any obvious signs of overheating.

Some specifics:

  • The CCS ICs for V1 and V2 are apparently fine; I can adjust both to a stable 175 V.
  • V1 and V2 themselves are behaving normally.
  • Adjusting bias pots R12 and R23 has little or no effect on output current.
  • B+ for V3 and V4 is 185 V (measured between ground and plates).
  • V3/V4 current was 167 mA each (!) as measured across R18/R29
So I have several questions:

  • First, I'm fairly certain I've blown the MOSFETs, but why am I not reading a short between pins 1 and 2 (drain and source, if I've read the pinout correctly)?
  • Also, around several of the board's mounting holes I noticed that there is very little clearance between the holes themselves and the adjacent traces. The #6 stainless washers I used in my mounting hardware appeared to be dangerously close to overlapping those traces. Is it possible I've created a short to ground? Do I need to use insulated washers or other hardware?
I've already filled out a shopping cart at Mouser for lots of spare silicon, but I'd just like to know if I've overlooked anything before hitting the "Order" button (I already know better than to ask for a check on my sanity :D).

I found it odd how the amp continued to produce somewhat decent audio after the magic smoke escaped. There was a noticeable increase in distortion and a drop in output, but I couldn't hear anything else dramatically obvious. But what really amazed me was how those cheap 2A3B's (more like 2A300B's) ate 30 W each without a burp. Not even a hint of red plating during the time it took me to realize what had occurred - which was embarrassingly substantial.

The magnetic components, however, were very unhappy. They told me not to let it happen again.
 
Last edited:
  • Like
Reactions: 1 user
The first question is "what smoked?"

Remove the output tubes and measure the DC voltages. Despite the specifications stating their surge withstanding capability, another builder had an R5 (the 270 ohm in the negative supply) fry on power up. The negative voltage supply needs to come up before the B+ supply to keep the output tubes cutoff until they are warm, so the negative supply is solid state rectified. R5 limits the current and widens the conduction time to reduce noise. It however sees a large current pulse on power up to charge C6. The Vishay part listed in the BOM may not be up to the task. The circuit is the same as on the first gen TSE, but the old part number was no longer available at Mouser.

I have two working boards that have been abused with no issues, but that's two resistors. I plan to test the other 8 I purchased for the original build, and three compatible parts from other suppliers in the next few days. They arrived in a DigiKey box yesterday, but my wife has been in the hospital, so Tubelab activities have been near zero, the box is still unopened.

If R5 is dead, I would try the Yageo RSF200JB-73-270R which is most like the part in the original TSE, or the Ohmite 42J270E which is wirewound, and usually capable of eating such a large pulse.

If R5 is OK, check the negative supply, and then probe the grid pins on the empty 2A3 sockets and verify that the bias voltage can be varied over a range from about -35 to -100 volts. This voltage will vary as lot with different power transformers, but you should be able to hit about -50 to -70 for 2A3's. If this works, the mosfets are fine.

how those cheap 2A3B's (more like 2A300B's) ate 30 W each without a burp.

I am running mine at 380 volts B+ and 70 mA each current. That's 27 watts and they don't seem to mind. They are 10 year old cheap Chinese units. I think they were under $20 each when I got them.
 
Thanks, George. I didn't think about R5, and I didn't perform the "tubeless" tests because the event occurred at night when I was tired and frustrated - precisely the wrong time to go poking about a high voltage circuit. I'll check it out tonight when I get the time. I'll still place an order for more "sand" though, mainly because I'd like to have some spares on hand "just in case".
 
Give the man a cigar...

...R5 was indeed the culprit. At George's suggestion, I examined it visually and was a bit disappointed - the resistor looked absolutely normal. But when I measured it, my DMM read 3.1M(!).

Thinking this surely must be an error, I broke out the opti-visors to get a closer look and sure enough, there was a small crack almost perfectly bisecting the purple color band. No charring, burned traces, melted solder - nothing. I've never seen a resistor fail like this before, and without magnification the thing looked so normal I never would've seen it here if I hadn't known to look.

I searched my parts bins and found a 250 ohm 5W cement resistor, so I soldered that in as it measured within 5% tolerance of the original. It works perfectly.

In retrospect, I'd been goosing the power switch quite frequently during fitting and testing phase, and that must've done it. Lesson learned. I'll still place that Mouser order to lay in a supply of spares and to safeguard against part extinction (another lesson learned).

Thanks to everyone for your input!
 

Attachments

  • Bad_R5.jpg
    Bad_R5.jpg
    649.7 KB · Views: 1,008
These two failures have led me to discover that plenty of resistors are "suitable for high surge currents," "rated for high inrush current," or "surge capable," but virtually none actually have real specs for the amount or duration of such current. Some however have ratings for time and amount of overload duration.....like 5X rated power for 5 seconds or 10X rated power for 5 seconds.

This means I get to test a bunch of resistors with a BIG cap to see which ones can eat the most current for a few milliseconds.

The Mouser part number for R5 is being removed from the BOM. I will post a new one once I have blown up some resistors. Meanwhile a wirewound or metal oxide part is probably the safest choice.
 
When min blew I was using a CL-60 that just happened to be handy. I'm just glad I didn't have to rip everything apart to change out the MOSFETs as it's a fairly complex build, mechanically speaking. Sounds like you were able to get yours up and running well.

And you're right, it is an elegant, beautiful-sounding design. Mine is in the final stages of completion (90% finished, 90% to go - pictures coming soon), but it's going slowly. I can't stop listening to it long enough to get any work done!

That's a good place to be. :)
 
Hi quinine,

Yes, the CL-60 still in place. It seems to work well, though boy does it get hot! It gets plenty of airflow however, and it's well away from anything even remotely flammable.

I tested my amp with 45's for a short while, and loved them. To help maximize their lives I ran them at around 30 mA current, which yielded ~9.5 W plate dissipation. I'm just glad that resistor didn't cook off while they were installed; the resulting current surge would've made them extremely unhappy. The "2A300B's" didn't care and are still going strong.

I absolutely love this amp, and am also considering a 300B version. My gut tells me to wait and see how the UNSET plays out, though. It sounds intriguing, and I already have the iron - although it's difficult for me to imagine anything sounding much better than this.
 
Last edited:
Help replacing R5

First of all i sorry if my english si not so good.
All was fine until R5 flashed. Quickly i switched it off, but i don't know if another part got broken. i unsoldered R5 and it's burned (1.9Mohm).I was using Cl-90.
I don't know if i must use the Ohmite suggested in post 2, or two 540ohm in paralell, or another suggestion. What do you suggest me??
Using 300B, Trf: 650v, 5AR4, B+:360v, Bias:70mA.
Thanks!
 

Attachments

  • IMG_20190922_132753.jpg
    IMG_20190922_132753.jpg
    1 MB · Views: 488
Last edited:
Sounds like you're on the right path. Until the results are in on George's resistor "cook-off", I'd try the Ohmite. As I noted in post 6, I was able to shoehorn in a 5W cement resistor, but it was a tight squeeze - and probably extreme overkill. But if you have one of those handy from ~220 to 300 ohms, I'd say use it. :)
 
Member
Joined 2015
Paid Member
This just happened to me too. This is my 3rd TSE, and I didn't have an issue in the first 2 builds.
I missed the memo and ordered parts off of the old version of the BOM and had the PR01 part. Mine is visually burned, compared to yours.
The irony here is that for the first 2 builds I used boutique/audiophoolery parts (I think was either a Takman or TKD). Those parts were physically much larger than the PR01 parts and had no trouble at all.

I found it odd how the amp continued to produce somewhat decent audio after the magic smoke escaped. There was a noticeable increase in distortion and a drop in output, but I couldn't hear anything else dramatically obvious.
Same here, on initial start-up everything checked out.
I shut the amp down and moved it over to the area where I have my PC based spectrum analyzer to get measurements. This is a 300B build, and I was expecting around 7W output based on the operating point. At 1W output, 1% distortion, and close to 10% distortion at 3W. This was much lower than expected, so I started troubleshooting and noticed that the 300B's were at 150-200mA bias !!! This dropped B+ to around 280V though, so plate dissipation didn't exceed that much.
 
Until the results are in on George's resistor "cook-off", I'd try the Ohmite.

I bought a bunch of resistors, and no surprise, couldn't get any of them to blow up when I tried even with an oversized power transformer resulting in 450 volts of B+ (no output tubes). The only resistor that I could kill was the Vishay PR01 part mentioned in an early BOM, still I had to work at it and only blew one. A few users have seen this part blow, but many have not. I still would not trust it though.

I have been using 270 ohm "270W-1" parts that I bought 100 of about 15 years ago from Digikey in all my builds because I have lots left. I did buy some of the newer "270W-1" parts from Digikey for the test, they do look a bit different, and none blew up, but it would be wise to err on the side of caution and use a 2 watt part.

The resistance value is not critical, but this part sees the current surge that results from charging a 100 uF cap right off the transformer. A resistance is needed to reduce the peak ripple currents reducing power supply noise. Too low a resistance results in more noise, and too high a resistance results in excess dissipation, especially in the first few seconds after switch on. 200 to 300 ohms is probably OK.
 
Thanks George, that makes sense and will no doubt be valuable to those contemplating future builds. I figure if you haven't been able to blow one up...

Regarding that capacitor-related current surge, I was rather wondering if the reverse could be true; that the 7H choke I included in the power supply helped induce a spike on shutdown that cooked the resistor. I haven't reviewed the other thread to see if there was a choke involved there. At all events I'll probably keep that "overkill" 5W resistor in place for the foreseeable future. The amp has been working beautifully for 2-1/2 years now and even survived a trip to Burning Amp 2019 as carry-on luggage, so no sense poking that bear. I have too many other things to break. :D
 
Last edited: