Adding diodes in series

[locky]

Another question on the PSU theme...I presently have +-39vdc and want to reduce this to +-30vdc by adding in some diodes in series (A voltage regulator circuit seems a bit complex). My question is do I have to place the diodes on both rails of the circuit or can I simply place them on the positive rail? If I go for the latter then I presume I will have less voltage on the positive rail than the negative. Is this the case?

 

[LiloLee]

I'm sure diodes will not reduce voltage.

Depending on your circuit, if it uses a voltage regulator, you may be able to reduce it there, or you need to use a high wattage resistor in series. Now don't ask me what value because I was never very good at calculating things like that.

 

[T-bone Sanchez]

I presume he means zener diodes used after a resistor, I think they only go up to about 8v so he's looking at the series way. Thats what I think anyway.

I would presume you'd need them on both rails.

 

[ChrisPa]

I'm sure diodes will not reduce voltage.

Diodes will reduce voltage. There is a forward voltage drop of about 0.6V on a normal silicon diode, so if you put 10 of them in series you wll get a voltage drop of ~6V - 39V would drop to 33V.

In answer to Locky's question, yes you would need it on both rails to maintain symmetry, but...

Gut reaction is don't do it - the 'complexity' of a regulator should be compensated by a much higher quality (perfomance) PSU - the multiple diodes will still leave you with an unregulated supply with associated ripple and lack of performance. Bite the bullet and build yourself your first regulator. Simple 3-pin chip regulators aren't hard to implement

 

[Graham C]

Also, 10 diodes is 19 more dodgy solder joints and points of failure

 

[locky]

I hear what you're saying...can anyone point me to a simple 3 pin regulator that would do for 39v per rail at 8-8 A ?

 

[LiloLee]

Diodes will reduce voltage. There is a forward voltage drop of about 0.6V on a normal silicon diode, so if you put 10 of them in series you wll get a voltage drop of ~6V - 39V would drop to 33V.

Well I was nearly right and I'd need loads to drop the voltage in my valve amp

I would some implementaion of a LM317 http://www.national.com/ds/LM/LM117.pdf

Hang on, I just re-read your post. Is that 8 Amps? Wow what are you trying to weld The 317 is only good for 1.5A

 

[locky]

Let me give you the full picture...I have a transformer which is giving about -+ 28vac at 165VA which makes about 6 amps. This is being rectified by a bridge diode rectifier to give 39 vdc once it has been smoothed out by the capacitors. This is a little high, as I need 35vdc maximum. I need to reduce the dc voltage to +-35vdc. Does anyone know of a regulator that will handle this kind of current?

 

[themadhippy]

if you need that kind of current then your looking at making regulator out of power transistors,have a look for amplified zenner circuits,the zenner sets the voltage on the base of a power transistor.another dirty way is to reduce the A.C using capacitors, droping 4 volts at 8 amps is 0.5 ohms,so a 0.00636 farad capacitors in series of each power rail should work,dont use polarized caps though

 

[LiloLee]

Dawned on me that we are looking at this the wrong way around.

What is it you plan on powering? I ask this because the xfr may be able to produce 6A, but of course you may only be power a DAC for instance which is unlikely to draw more than a few 100mA's.

And BTW I still go for the power resistors to cut the voltage if you need high current. I think the calculation is
(Initial Voltage - Required Voltage)/Current, so (39-35)/6 = 0.7ohm which needs to be rated at 16watts for 4A, 12w for 3A etc. CPC do a 12w 1R which may do.

BTW this is what TMH was talking about http://www.allaboutcircuits.com/vol_3/chpt_3/9.html

 

[zanash]

Lots of diodes will add noise into the system....not good
Big power resistors will add/ affect the sound produced by the psu minimally
Power transistor will also add noise but no more than any other voltage regulation.
I 've not checked so could be wrong but the must be a myriad of voltage regs. suitable to drop 8vdc at 3amp....

Though I'd be tempted by the simple resistor approach.

 

[wadia-miester]

If you go that route, then use a caddock 1% thick film unit (with heat sink say 6.8C/W) aint heard a better off the shelf unit, ultra low noise about the £6 mark not cheap, but then good stuff never is. Wm

 

[NRG]

As mentioned what is the current draw of the device you are powering? The LM338 is rated for 5A and 7A peak.

 

[LiloLee]

Of the top of my head, the LM338 can't handle 39v input voltage.

 

[NRG]

The input / output _differential_ is 40v max.

http://www.national.com/ds/LM/LM338.pdf

 

[locky]

Thanks for all the ideas...I'm tempted to go down the power resistor route. Does a 1ohm 30W resistor plus heatsink sound reasonable? I'm presuming that I will need one for each rail? By the way, the psu is powering a D-class amp from 41hz.

 

[LiloLee]

Doh I'll check the datasheet and not my memory next time.

So it can provide up to 32V, which means up to 72V input voltage.

But locky what are you building and what is the current draw?

 

[locky]

It's to power a DIY amp module. See http://41hz.com/main.aspx?pageID=88
I'm not sure what the draw of the module is, but the amp needs a 30-35vdc supply as well as a +5vdc supply.

 

[wadia-miester]

The 5v supply is for the dsp chipset and needs to be as clean as possible

 

[NRG]

If you are going to drive 50W into 4R with +/-30v rails a 300Va TX is the minimum needed. Ia peak is 5A into load, I would regulate the supply with the LM338. Heatsink well.