Installing SS Rectifiers for Tube Rectification

[Gtr0]

Hi Folks.

I have read the various posts here and there on adding diodes in series before tube rectifiers to protect the surrounding circuit/keep playing etc should a tube rectifier blow out...

I made up a few drawings of "practical use" (attached), could one of you fine wizards confirm (or other) that this is correct before I install them real world?

Thank you SOOOOO much!!! Fingers crossed I get this right.

[pdf64]

The schematic is correct but if they are called into use by tube failure, 1kV diodes are underrated for HT windings much over 600V.
Need to factor in highest mains voltage (eg x 1.08) and unloaded HT (eg x 1.1), then the rms-to-peak ratio x 1.414. So that works out to ~1.68 x the (loaded) HT winding voltage; so ~1k2V for your PT.
UF diodes are of no benefit in regard of switching noise (until a 5U4 shorts), as the conduction characteristics are defined by the 5U4 (it will switch off way before a silicon diode).

I don't like how you've drawn the 5U4 plates already shorted together!

[Gtr0]

The schematic is correct but if they are called into use by tube failure, 1kV diodes are underrated for HT windings much over 600V.
Need to factor in highest mains voltage and unloaded HT (eg cold tubes), then 1.414 x that HT winding voltage; so ~1k2V for your PT.
UF diodes are of no benefit (until a 5U4 shorts), as the conduction characteristics are defined by the 5U4.

I don't like how you've drawn the 5U4 plates already shorted together!

Perhaps a BY227 diode then? 1250V, 1.5A

How else could I connect the plates ?

Thanks, PDF64!

[pdf64]

Yes, it's not critical, just need a reasonable degree of voltage and current 'headroom', so 1250V 1.5A should be fine.

With a directly heated cathode like a 5U4, the cathodes and heater are all one thing, but the plates are independent, only linked via the HT winding see

[Gtr0]

Yes, it's not critical, just need a reasonable degree of voltage and current 'headroom', so 1250V 1.5A should be fine.

With a directly heated cathode like a 5U4, the cathodes and heater are all one thing, but the plates are independent, only linked via the HT winding see [ATTACH=CONFIG]54751[/ATTACH]

Excuse my ignorance here, the only other way I can think to do this is as such...

[Tony Bones]

Two 1N4007 in series (for a total of four) will give you the PIV headroom you need. Not as easy to mount on the tube socket though.

[mikepukmel]

Oh boy did I misunderstand RG's 'immortal mods' article. What I don't understand: so suppose you have a 345 - 0 - 345 vac rms secondary power transformer. The center tap is grounded. When charge moves in one direction, one diode (I thought) only sees 345 * 1.414 = 488v dc peak, and this is reverse biased. When this diode is reverse biased the other one is conducting, and (I thought) the conducting diode sees 488v dc peak fwd. I thought that since the center tap is grounded, each diode only sees current from half the winding, since the other half is "off" because its diode is reverse biased. Where did I go wrong?

[pdf64]

Oh boy did I misunderstand RG's 'immortal mods' article. What I don't understand: so suppose you have a 345 - 0 - 345 vac rms secondary power transformer. The center tap is grounded. When charge moves in one direction, one diode (I thought) only sees 345 * 1.414 = 488v dc peak, and this is reverse biased. When this diode is reverse biased the other one is conducting, and (I thought) the conducting diode sees 488v dc peak fwd. I thought that since the center tap is grounded, each diode only sees current from half the winding, since the other half is "off" because its diode is reverse biased. Where did I go wrong?

Have a read of http://www.valvewizard.co.uk/bridge.html
The way I see it, on the negative going swing of its HT winding leg, the diode's anode will get pulled down to -Vp whilst its cathode is up at the HT Vdc; so pretty much twice the rectified HT Vdc across it.

By writing the the 1N4007 was good as a back up for HTs up to ~450Vdc in his immortal amp mod article https://www.premierguitar.com/articl...mplifier-mod-1 I think RG may have oversold it. Though in reality, 1N4007 generally seem very robust, and able to accommodate reverse voltages somewhat above their rating.

Excuse my ignorance here, the only other way I can think to do this is as such...

The symbolic representation of the rectifier tube in your schematic of post #1 shows that the pin 4 plate is internally connected to / part of the the same 'electrode structure', as the pin 6 plate, ie so it would short the HT winding. That's not how it is, clearly; the pin 4 and pin 6 plates are separate, electrically isolated structures within the tube envelope. Hope that makes sense?

[Helmholtz]

I think it is easier to understand max. PIV (peak inverserve voltage) across the diodes if you ignore the CT and ground reference as they don't change the voltage across the diodes.
The RMS voltage across the full secondary is 2xVrms, Vrms meaning the voltage of one secondary half.
During voltage peaks the blocking diode sees the full secondary voltage as its cathode is connected to the conducting other diode (which may be considered a short) while its anode is connected to the other end of the full secondary. So the diode is effectively connected across the full secondary and its PIV is 1.414 x 2 x Vrms.

As Merlin explains it is not safe to assume that inverse voltage is equally shared between diodes in series. He recommends to add parallel capacitors.

[Jazz P Bass]

"As Merlin explains it is not safe to assume that inverse voltage is evenly shared between diodes in series. He recommends to add parallel capacitors. "

Wouldn't a high value resistor (10Meg) across each series diode do a better job of current sharing ?

[Helmholtz]

As the voltage across the diodes is pulsed, capacitors > 10nF will do the voltage sharing job. Merlin also mentions the option of using high value resistors, but 1kV rated caps are much easier to find than 1kV rated resistors.

[Gtr0]

The symbolic representation of the rectifier tube in your schematic of post #1 shows that the pin 4 plate is internally connected to / part of the the same 'electrode structure', as the pin 6 plate, ie so it would short the HT winding. That's not how it is, clearly; the pin 4 and pin 6 plates are separate, electrically isolated structures within the tube envelope. Hope that makes sense?

If I understand correctly, then we are talking about a mistake in my schematic drawing of the 5U4GB, right? For sure I see the error there now. What about the illustrator drawing of how I will physically connect them? I have the feeling that I am missing something painfully obvious here.

As the voltage across the diodes is pulsed, capacitors > 10nF will do the voltage sharing job. Merlin also mentions the option of using high value resistors, but 1kV rated caps are much easier to find than 1kV rated resistors.

You usually see 8200pF caps in Fender's amps... they seem to be the only ones doing this, or at least that I have seen. I am glad this came up, reminds me that I have been meaning to seek out some 1kv caps for this reason.

[Gtr0]

Now I remember how the 10nF caps fell off my radar, my usual suppliers (TAD and Tube-Town) never carry them, but I recently found a few new ones (Modulus Amp and Banzai) that had them, so I ordered some of those. I suppose I will need to move the diodes and caps to a more secured locations (off socket).

[Helmholtz]

I suppose I will need to move the diodes and caps to a more secured locations (off socket).

Or use single sufficiently rated (1250V/1A or higher) diodes like BY127/227/133/255 without the caps.

https://www.donberg.fr/catalogue/sem...bz/page_2.html

[trobbins]

I would recommend using 2x 1N4007 for each 'ss' diode, without any additional balancing parts - with the proviso that your 1N4007's were bought as a single batch of new parts (ie. not the dregs from a parts drawer that has seen lots of diodes come and go). There is inherent balancing of PIV if any leakage from one ss diode starts to occur, but the PIV margin is going to huge anyway. Any thought of adding extra balancing parts is not warranted imho due to mains frequency application, and only using 2 diodes in series (rather than say >10), and as an aside it is very impractical to install 1kV rated resistors (eg. 3x 350V 0.4W) and/or capacitors.

Keep the valve socket clean if you are going to utilise the spare pins - I appreciate they look tantalising simple to use, but the working anode pins are spaced for a reason.

I would also recommend using one 250VAC fuse in the CT, rather than 2 in the secondary arms, as there may be collateral damage if one fuse fails. Also, try to select a fuse rating and type that achieves the best discrimination possible - as although the fuse may not blow in time to save collateral damage, it is likely a key part in avoiding damage if it is an appropriate value and type.

There are advantages in using the valve diodes in each valve 'in parallel', rather how you are using them, but there are also disadvantages. At least with your use, you can remove one 5U4 if needed, and you will still get B+.