Replace Selenium Rectifier with Silicon Diodes & MOSFET Voltage Dropper in Tube Radios

[@matambale]

Again, very smart design (downright clever!), clear explanation, thank you Manuel.

[@electronicsoldandnew]

Pleasure

[@SimonB6706]

I’m just refurbishing a Grundig Mandello GB and decided to try this to get B+ at the correct voltage. The PCBs have arrived this morning so as soon as the MOSFET arrives I’ll get the job finished. Thanks for this 😊

[@electronicsoldandnew]

My pleasure. By the way, I’m in the process of adding it to a new project which should be posted this weekend, and it works perfectly 👍

[@SimonB6706]

Following up, I've built the circuit and fitted it to the Grundig Mandello GB. Initially it didn't function as I was just getting the full silicon rectified output of ~300Vdc. The fault was because I'd used metal standoffs and the one near the C2/C4 pads shorted those and hence the anodes of D3/D4 to ground (fortunately!). I'd suggest dropping the C2 - C5 caps and using the space for a proper mounting hole pad on the next version of board. Now it's working I can successfully adjust the voltage to the 292Vdc spec. Thanks again.

[@techobaz55]

A very simple, smart design Manuel. Will solve many issues I feel, beyond the Selenium application as well ! Thank You !

[@electronicsoldandnew]

Pleasure

[@JonRig]

Yet another slice of EO&N genius!!! Excellent work and really great having the detailed description. Well done Manuel!!

[@electronicsoldandnew]

Thanks

[@ibrahimkocaalioglu]

Good design. Hope it doesn't fluctuate when current draw changes.

[@electronicsoldandnew]

It doesn’t. Only when the mains supply changes slightly, it obviously changes a little.

[@gang208]

I believe a serial resister (as you did in Telefunken Opus 8 restoration) can actually work slightly better than a zener diode (or its MOSFET equivalent). The resister with the capacitor forms a 1st order LPF that reduces the ripple in the output. There is a rush of current at the moment the zener is turned on. A resistor limits and smooths out the rush current. You can verify this with a Spice model.

[@electronicsoldandnew]

👍

[@tubeDude48]

Another EXCELLENT EO&N design! 👍 Thanks, Manuel!

[@electronicsoldandnew]

Im sure someone has done this somewhere before, but I couldn’t find it.

[@tubeDude48]

@@electronicsoldandnew - I've not come across this as well! 👌

[@SBWings]

I'm just in the process of placing my 1st order with PCBWay. Thanks for this!

[@electronicsoldandnew]

Pleasure

[@SBWings]

@@electronicsoldandnew Choosing the black solder mask increased the production time slightly. This turned out to be fortunate. I was able to add two more of your projects, your capacitor leakage tester and the capacitor outside foil checker. Since the selenium dropper was still in production, PCBWay allowed me to combine shipping. All three projects will ship at the same time!

[@electronicsoldandnew]

👍

[@ericrawson2909]

Clever design. I was watching for the voltage to change as the FET warmed up, but it didn't. I thought, doesn't the Vgs/Drain Current characteristic change with temperature? I looked at the datasheet for NTP067N65S3H which I recently used, and for a given current, Vgs changes by as much as half a volt between 25C and 150C. This would be magnified by the potential divider ratio of the trimpot. I guess it's more than accurate enough for a valve radio supply, and a very economical design. Edit: by the way, I used Pcbway on your recommendation and can recommend them. Quick delivery and perfect product quality.

[@electronicsoldandnew]

The temperature will affect the response, but, as you said, tube gear is very tolerant.

[@anandarochisha]

Brilliant work, Manuel. You could be mistaken for a very talented Electronic Engineer. 😊

[@electronicsoldandnew]

Mistakes are made all the time 😊

[@anandarochisha]

@@electronicsoldandnew if you are like me, the more mistakes I make, the less money I have to spend on hair care products.

[@electronicsoldandnew]

😊

[@vcdios]

Thanks a lot for the video, I’ve already ordered the PCB. I read about this method in Radiomuseum, but I still find your method of winding the primary extra winding more interesting because it adjusts the entire secondary voltages. In this sense, what do you think about adjusting the heater voltage with a series circuit using a capacitor to fine-tune the voltage?

[@electronicsoldandnew]

If I need to drop a little from the heater voltage, I just normally use a very low value 2W resistor. It works just fine.

[@BKGoldDetecting]

Another great design, I’ll try this on my next radio. Incidentally, do you do anything to address the heater voltage being too high when you modify a 220v radio to 240v. I’m getting close to 7v on my Freiburg even though I’ve got the B+ correctly adjusted.

[@electronicsoldandnew]

I’ve had that issue before, and used a resistor (very low value, like 0.22 or 0.33 Ohms) in series before it goes to the tubes. I used one of those heat-sink 10W ones and it worked brilliantly

[@fredrikviklund5067]

Great idea and well explained and executed as always! One thing I'd love to see in a follow-up is a failure mode analysis. What happens if the MOSFET fails short, if the pot is shorted etc. Are there any circumstances where this circuit could add a new risk or is not suitable? A systematic discussion of any potential issues would be a great addition.

[@electronicsoldandnew]

The failure modes are quite straightforward: shorted or open. If it opens, the circuit is simply cut off as if you’d used a switch to switch the supply off. If it shorts, the original B+ will result with no drop. The circuit should be able to. Andre the extra 10 or 20V with no major issues.

[@poormanselectronicsbench2021]

Nice little board design! One thing you can do to create some more heat dissipation for the MOSFET is, move the AC2 and + output through holes closer to the diodes / trimpot area, then, do a copper fill on both sides of the PCB , and then drop some 1.5 ~ 2MM via's outside of the MOSFET's mounting area to create some more cooling surface area. You would just have to make sure that the mounting tab area does not contact anything else when mounted, but according to the Vishay datasheet that tab is not shown as connected to the D-S or G pins.( If you say it is a DRAIN connection, then no harm or foul as it will be chassis potential anyway, as you stated)

[@electronicsoldandnew]

👍 I placed the holes for AC, + and -, in their positions to match those on the most common selenium rectifiers, but I see your point. I checked on the tap to D connection, and it is indeed there.

[@poormanselectronicsbench2021]

@@electronicsoldandnew The Vishay datasheet for that particular MOSFET does not clearly state the tab is in fact a Drain connection on the lead designation drawing, but, going to the "Thermal resistance Rating" section of the chart, it does list a value for a "Maximum Junction to case (Drain)" , which round-aboutly infers that the tab is in fact a Drain connection, odd it was left off of the lead drawing info. Like you said though, since it is at ground / chassis potential in most applications, there's no harm or foul in also having a larger copper area connected to it if you do expand it on the PCB.

[@electronicsoldandnew]

👍

[@Phil-Webb]

You said that you couldn't add the circuit to your Braun radio. Would this be an improvement, or is it not possible to replace a half-wave rectifier with a full-wave one?

[@electronicsoldandnew]

I could replace it but I don’t want to change the circuit too much from its original condition. In restorations, I try to keep it as original as possible.

[@alainblazit9504]

A MOSFET and a Zener can make a voltage regulator, to avoid B+ overvoltage during startup.

[@goosgitaar]

Gone try it Out ! Had no succes with the zener version a terrible hum apeart In the guitar amp

[@bofor3948]

I like the design of this as it simplifies the adjustment of the B+ difference and it can be setup with a lower input voltage. This does of course bias the whole chassis above earth and in some cases this can interfere with the bias setups already exisiting, so this would be unusable. Also you would have to be careful not to add a mains earth or ground earth to the chassis. I can see the point about about the risk of high potential contact if placed in the B+ line, but if needed could this be potted (allowing access to trim pot) or non metallic bolts used to reduce contact risk? I know if extra heatsink is required, this would rule out the chassis.

[@electronicsoldandnew]

If the radio doesn’t have any bias arrangements around the ground system, and if the chassis is connected as a general ground, earthing the chassis shouldn’t be an issue. This can just as easily be used on the hot side, with some extra precautions, as you mention.

[@mackfisher4487]

Manuel, I want to add to the Praise of others. If PCB way marketing department is reading the comments might I suggest they send you a Christmas bonus.

[@electronicsoldandnew]

😊 Bonuses are always welcome 😊

[@georgebliss964]

I like the simplicity of the series zener diode. To reduce the wattage of that single one, would 3 x 6.8V or some other combination in series do the trick?

[@electronicsoldandnew]

Yes, you can add zener a in series.

[@hahaagi]

FET GATE 와 SOURCE 사이에 PROTECTION ZENER DIODE(20V)를 넣는게 더 좋지 않을까요?

[@electronicsoldandnew]

Not necessary here as there are no caps on the negative side that would cause the G-S voltage to go too high.

[@randomsteve4288]

A GS protection diode may be added (15V preferably) but should not be needed. Due to the feedback constellation the Gate Source voltage will float at the Gate Source Threshold Voltage. Any rise in Gate Source voltage would cause the MOSFET to conduct more, which drops the DS voltage, and that in turn cuts down on the voltage to the GS divider. The MOSFET protects itself from too high Gate voltage. This circuit is just a MOSFET version of the well known "adjustable power Zener" also known as "Vbe multiplier" using two resistors and a bipolar Transiator and which has been around since the dark ages.

[@murilomonconill6229]

Perhaps for diodes D1 to D4 microwave oven diodes could also be used, such as the HVR-1X-4 or similar, which has a voltage drop of 10V and a current capacity of 500mA. No additional circuit would be necessary, except that it would not allow linear adjustment of the voltage drop.

[@cbiz8]

Hi, does the voltage drop in the MosFet remain constant by varying the load current?

[@electronicsoldandnew]

It does, as is seen by the increasing voltage across the same load (hence increased current).

[@cbiz8]

@@electronicsoldandnew Ok, gracias!

[@DavidSmith-zx7wz]

How did you get so smart! Are you self-taught, or ? Thanks for what you do.

[@electronicsoldandnew]

😊

[@wolfgangrichter6088]

Why not put two thyristors into the rectifier bridge instead of diodes and fire them in a way so that Ub will be at the voltage you like ? Of course it will introduce some switching noise, but maybe it is not as bad as one could fear because you switch the thyristors on at a low voltage already and if you shield and filter it properly with a few ferrites here and there, the radio may not be influenced terribly. HP used this in it's high wattage lab power supplies before high frequency switchers came into use, if I remember right.

[@electronicsoldandnew]

That would make the supply a little more complicated than it needs to be 😊

[@wolfgangrichter6088]

@@electronicsoldandnew : To start with, you could use an AC dimmer. Thats one trimmer, one condensor, one diac and a triac.

[@colingymer]

Am I missing something Manuel, you do not appear to have specified the MOSFET?

[@electronicsoldandnew]

I did on the schematic. IRF840 or similar.

[@colingymer]

Thanks Manuel, missed that...

[@BersekViking]

I see one problem with this circuit. If you turn the potentiometer all the way down, the mosfet never turns on, and the full voltage will be over the potensiometer. The potentiometer will then dissipate about 2 watts. You should add a fixed resistor at the bottom of the potentiometer to reduce the maximum voltage allowed over the mosfet.

[@electronicsoldandnew]

That’s a good point

[@BersekViking]

@@electronicsoldandnew While you're at it, I suggest you out put a high value resistor, like 1M, between the top of the potentiometer to the whipper. The will prevent disasters if the potentiometer gets oxidized and the whipper loses contact with the track. 🙂

[@electronicsoldandnew]

👍

[@sylweksliwka7986]

Super

[@electronicsoldandnew]

👍

[@opera5714]

R1 serves no useful purpose. Use that space to put a resistor in series with the FET to dissipate some of the heat into the resistor instead of the FET.

[@BobAndersson]

Interesting. I may be missing something but I don't think this would work on the Saba Freudenstat I did recently as that is one of those radios where the negative output from the bridge rectifier is connected to chassis ground via a couple of resistors to provide a grid bias potential. BTW if the MOSFET tab can be electrically connected to chassis it "could" be mounted underneath with a spacer between it and the board and thus be bolted direct to the chassis if cooling ever did become an issue? Personally I think I'll continue to use a large bridge rectifier (usually a Comchip GBPC3510-G which doubles as a tagstrip!) and a power resistor which can both be safely bolted to the chassis. That also allows for an extra 50uF capacitor placed immediately after the rectifier to add an extra (probably unneeded 🤣) stage of pi filtering, if that's a phrase? But variety is the spice of life and thanks as always for sharing Manuel. ⭐

[@electronicsoldandnew]

That Saba needs the bias, so it probably won’t work as it’s presented here. Also, those SABAs draw a LOT of current, which will really demand a good heatsink solution.

[@FBord1139]

I understand the need for a voltage drop, but see two issues with a resistor only circuit: startup current vs voltage drop and dependence on the turn-on curve of the mosfet. Suggest adding a TL431 (cheap $0.30) as an active zener in your circuit. Lookup 'Variable power zener circuit using tl431.". It will achieve a constant voltage drop independent of startup current, temperature and mosfet characteristics.

[@electronicsoldandnew]

Good idea. I’ll look into that.

[@raymondmenard5444]

Wow ! This is just brilliant Manuel and would you believe i was just looking at solutions for replacing the AEG rectifier on the Grundig 2220CA i'm working on at the moment ? Of course i will still have to build a new case in order to adapt it to the chassis and fit this board, but at least adjusting the DC output will be much simpler. I have used your other rectifier board on a Blaupunkt Riviera 2540 USA i just finished restoring and it has been working flawlessly, although i find the diodes while not going out of max specs do heat up quite a bit. I wonder if this is normal ? Anyway for those interested in seeing my rectifier replacement and if you allow it Manuel, you can watch it in the final episode of the series (part4) starting at 18:45m. kzbin.info/www/bejne/e2W7ZGd_i7B2r6c

[@electronicsoldandnew]

👍 the diodes do get a little hot. That’s normal. I have no problem with the video showing the boards. 👍

[@raymondmenard5444]

@@electronicsoldandnew Ok great, thank you Manuel. I sort of thought the diodes' heat level wasn't too too bad, but i just wanted to make sure with you. Thank you for allowing my footage, trying really hard to start my own channel, but man i'm really more a photographer than a cinematographer. I don't know how you guys do it on a daily basis :-)

[@electronicsoldandnew]

I only do it on a weekly basis, and it took me quite a while to get something acceptable 😊

[@gamiwv]

Very inefficient design. Learn SMPS.

[@johnstone7697]

This is only meant for replacement of a selenium rectifier bridge in a radio. The radio already has its own power transformer, thus your comment about SMPS is irrelevant. .

[@EuroScot2023]

If you want efficiency, you should tell Manuel to replace the entire chassis with a semiconductor one. That'll really lower the power consumption. Did you recently cook your head in the microwave?

[@milantrcka121]

Ignorant troll

[@TheEmbeddedHobbyist]

Looking at the circuit the fet is shown wrong. The body diode will be foward biased all the time.

[@electronicsoldandnew]

Nope. Look again, I know I had to 😊 The B- is actually lower in voltage than the GND point shown. I had to blink a few times myself to check, because it’s a little counter-intuitive, but I believe it’s correct, and is exactly as it was built and tested.