Breadboarding Poindexter's 6GK5/EL34 Music Machine

[boywonder]

I have just finished assembling Poindexter's 6GK5/EL34 design as a breadboard, and I am struggling to get past the white-knuckle, plug it in the wall and fire it up phase.

I'm using the PS design shown in the attachment, and so far, have powered it up through a variac.

The basic issue is lots of B+ overshoot on power-up that's not reflected in the PSUD model. Upon power-up, with a mains voltage (set on the variac) of around 80VAC or so, the B+ voltage peaks around 450V or so before settling down to the 380/320 values as expected (I crank up the variac to 120VAC after the B+ has settled). This surge lasts a good 30 seconds or so, and I imagine that if I let the PS power up on 120VAC that the B+ would exceed 500V+ for the 20-30 seconds or so. So I'm concerned that I will pop the TS-HA/HB caps with this overvoltage.

Could this be ringing due to the 2 L/C sections? It's not reflected at all in the model.

I also modeled a 5AR4 CT PS, but I appear to be on the ragged edge of the forward current rating of the 5AR4, so I started with the the SS design.

[dsavitsk]

The tubes take a while to start conducting. So, rerun the PSUD model with 0 current draw and see what happens to your voltage.

[boywonder]

Thanks dsavitsk! Simple enough, that explains it. Changing both current taps to 1 ma ends up with around 540 volts.

Sounds like heating the tubes before applying B+ will solve my newbie dilemma.

So I suppose that I need a separate filament transformer and a 30 second B+ transformer delay, or delay the B+ with something on the B+ CT while still utilizing the filament taps in my primary transformer.

If I use a 5AR4 design will I still need to use some type of delay? It seems that I will since it will bring up B+ in about 10 sec or so and I need 30 seconds for warm-up.

[Miles Prower]

Originally posted by boywonder
Sounds like heating the tubes before applying B+ will solve my newbie dilemma.

When using a solid state PS for hollow state equipment, this is a good solution to the problem. With a project using a solid state PS, I separated the heater circuit from the HV circuit, and wired in an interlock so that the HV won't come on if the heaters aren't energized. (This not only helps prevent applying HV to cold VTs which will ruin them, it also prevents the possibility of having HV present without the usual indications: glowing heaters, and making sound.) You could automate that if so desired without too much difficulty.

So I suppose that I need a separate filament transformer and a 30 second B+ transformer delay, or delay the B+ with something on the B+ CT while still utilizing the filament taps in my primary transformer.

Your best bet is to use a separate heater xfmr. The problem with switching in the HV DC neutral is the high voltage that exists at the switch. This could present a user hazard. Given the capacitors used, you don't need a HV delay.

If I use a 5AR4 design will I still need to use some type of delay? It seems that I will since it will bring up B+ in about 10 sec or so and I need 30 seconds for warm-up.

If you substitute hollow state diodes, you already have your HV delay. Another project I did derived the positive rail from a 5U4GB. This type, being a DH, still provides plenty of HV delay that allows all the VTs (except for the 6BQ6GTB finals -- those enormous cathodes take a good long while to heat up) to warm up before substantial HV appears. This design also included a negative rail derived from SS diodes. The negative rail does soar, but that's OK: you want the bias to appear first. It comes down to the design nominal value once the current demand goes up as the finals heat and start pulling current.

A 5AR4 will give even more HV delay since it's an indirectly heated type.

[boywonder]

Thanks Miles. The two issues that I presently have with the 5AR4 design are:

1 Getting to the target voltages without exceeding the forward current of the 5AR4. I cannot seem to get below .75A on the PSUD sims, and I'm assuming that that's the limit. Dual mono's look do-able, but with double the current draw for stereo I'm struggling. If I am interpreting the PSUD error message correctly, I am on the ragged edge for forward current for the 5AR4. Please enlighten me if this is not the case.

2. As mentioned above, don't I still risk overvolting the caps if the rectifier comes up over 10 seconds or so and the output tube filaments take 30 secs? It seems at 10-15 secs I'll have full B+ on semi-cold output tubes overvolting my PS caps.

Attached is the 5AR4 PS. The error message says that the recitfier forward current is .78A

[Bruce Anderson]

Poindexter uses a human powered B+ delay (I just comleted the 6GK5/6V6 and I used it also). Its all on his website. Try it, works perfectly.
PS Love my/his amp!

[Tweeker]

Have the heater current for parallel 5AR4?

[boywonder]

Originally posted by Tweeker
Have the heater current for parallel 5AR4?

I do. Can I just split the transformer B+ to a pair of 5AR4's, each with their own CLCLC?

[Miles Prower]

Originally posted by boywonder
2. As mentioned above, don't I still risk overvolting the caps if the rectifier comes up over 10 seconds or so and the output tube filaments take 30 secs? It seems at 10-15 secs I'll have full B+ on semi-cold output tubes overvolting my PS caps.

That is always a risk, and a very good reason to be generous with filter capacitor voltage ratings. When I used a 650Vct power xfmr with a 5U4GB, I made the first resevoir capacitor from two 68uF / 300Vdc units in series, to give the design 34uF total capacitance, and for a generous voltage margin. Since the HV could go to a maximum of 458Vdc, you can see why even though the design nominal voltage was 350Vdc. 450Vdc capacitors just might

Same treatment for the negative rail: 47uF / 300Vdc units in series. Electrolytics are cheap in comparison to what they may take out if they fail. When wiring in series for higher voltage, be sure to parallel each capacitor in the series string with 100K resistors to ensure voltage balance.

[boywonder]

Are there any downsides to using a pair of 5AR4's each with their own CLC feeding each channel?

I would assume that you would get twice the switching spikes in the transformer.