# Do I need a lossless resistor?

#### Xerox

Joined Aug 6, 2015
13
I need a dc-dc power supply whose output voltage tracks the input voltage with a fixed differential. For example I have 65v in and want the output to be 15v less at 50v. If the input voltages falls to say 50V then the output should fall to 35V and so on. The conversion has to be as efficient as possible into a fixed value resistive load of say 10 ohms. Any ideas anybody?

#### kubeek

Joined Sep 20, 2005
5,659
A buck regulator with a little more convoluted feedback than usual. What current do you need? Is your maximum 50V/10R = 5A or are those just some bogus numbers?

#### Xerox

Joined Aug 6, 2015
13
The numbers are real the current could go as high as 8A. I agree on the buck regulator. In ac terms it would be a transformer so really I'm asking for dc-dc transformer. As size is not an issue 100v electrolytic capacitors could be used in or around the input.

#### pwdixon

Joined Oct 11, 2012
488
Just stick a 15V zener in series with the load (could be too simple of course and not particularly efficient).

#### Xerox

Joined Aug 6, 2015
13
A resistor for any given current would do the same thing but little or no heat needs to be generated. I might just add that the dc input is not from a theoretical constant voltage source, output loading would increase or lower it.

#### AnalogKid

Joined Aug 1, 2013
8,254
Just stick a 15V zener in series with the load (could be too simple of course and not particularly efficient).
15 V x 8 A = 120 W

ak

#### pwdixon

Joined Oct 11, 2012
488
15 V x 8 A = 120 W

ak
In my defence my posting crossed over with the one before so I didn't know it was 8A when I posted.

#### Papabravo

Joined Feb 24, 2006
12,768
I need a dc-dc power supply whose output voltage tracks the input voltage with a fixed differential. For example I have 65v in and want the output to be 15v less at 50v. If the input voltages falls to say 50V then the output should fall to 35V and so on. The conversion has to be as efficient as possible into a fixed value resistive load of say 10 ohms. Any ideas anybody?
You may need a lossless resistor. Unfortunately such a thing does not exist.

#### PeterCoxSmith

Joined Feb 23, 2015
148
You could use something like a Linear Tech LT3790 and use the required differential between the input and output as the FB signal. The differential then becomes the controlled parameter.

#### GopherT

Joined Nov 23, 2012
8,012
Connect a 15v Zener to positive voltage rail and resistor to ground, feed the mid-point node to a PWM controller (LS494???) as the set point, Have the PWM controller drive a set of MOSFETs to power his 8amp load (may need a bank of capacitors) and fairly high PWM frequency to maintain the (Vss - 15V) target at 8 amps.

#### mcgyvr

Joined Oct 15, 2009
5,394
Maybe you should explain why you need a device like this?
Maybe we can offer "proper" suggestions at that point.

#### ronv

Joined Nov 12, 2008
3,770
Connect a 15v Zener to positive voltage rail and resistor to ground, feed the mid-point node to a PWM controller (LS494???) as the set point, Have the PWM controller drive a set of MOSFETs to power his 8amp load (may need a bank of capacitors) and fairly high PWM frequency to maintain the (Vss - 15V) target at 8 amps.
I think Gopher has the solution. This one is linear, but the same works for a switcher.

#### Attachments

• 50.4 KB Views: 9

#### Xerox

Joined Aug 6, 2015
13
Well that's the trouble if I did explain everyone would go off on a tangent and give a zillion reasons why it should be done another way. So I'll hold off on that for a moment if you don't mind. This is thinking outside of the box to challenge an established way of doing things. Lets say I have an electric shaver and that runs off 110v ac and I'm somewhere where the voltage is 220v ac and I need an electric shave. Simple, obtain an auto transformer and in a nearly lossless way it reduces the voltage for me. The shaver does not need a regulated output it will work with a wide range of input and output voltages. So why can't a dc-dc transformer do the same. My problem is that 65v is regarded as an unsafe voltage but anything 50v downwards is acceptably safe. So for safety reasons the dc voltage needs reducing in a highly efficient way on a par with an ac auto transformer. My problem is that the requirement is out side of the box but I'm still stuck inside.

#### Xerox

Joined Aug 6, 2015
13
I think Gopher has the solution. This one is linear, but the same works for a switcher.
How hot would it get?

#### ronv

Joined Nov 12, 2008
3,770
How hot would it get?
How hot would what get????
If it were a linear regulator it would get very hot. If it is a swither designed for 8 amps it would be fine.
In the circuit shown with the resistor shown the zener should be 1/2 watt. It will be ok.
The resistor should be 1 watt. Or you could make the resistor 5K then 1/2 watt would do.

#### ian field

Joined Oct 27, 2012
6,539
Just stick a 15V zener in series with the load (could be too simple of course and not particularly efficient).
From the current mentioned by the TS, you'd need a pretty big zener - like the ones supplied by Lucas for old British motorcycles.

Back in the 80s I looked into alternative sources, Philips listed 100W stud zeners the same physical size and in all the standard voltage increments.

#### Xerox

Joined Aug 6, 2015
13
"you'd need a pretty big zener - like the ones supplied by Lucas for old British motorcycles."

My very first attempt at doing this sent me to my junk box. Didn't have a big power resistor or enough stud diodes but had loads of 1N4001 so I paralleled lots of them and put them in series with some power stud diodes but things got hellish hot with the solder melting here and there. It was proof of concept but no way to go.

#### ian field

Joined Oct 27, 2012
6,539
"you'd need a pretty big zener - like the ones supplied by Lucas for old British motorcycles."

My very first attempt at doing this sent me to my junk box. Didn't have a big power resistor or enough stud diodes but had loads of 1N4001 so I paralleled lots of them and put them in series with some power stud diodes but things got hellish hot with the solder melting here and there. It was proof of concept but no way to go.
You can boost a regular zener with a heatsink mounted power transistor, power transistors are usually at the low end of gain spec, so you still need a decent size zener. Don't forget the transistor's Vbe drop when working out the voltage, if its not too critical use multiples of 1N4001 in series - if you need precise voltage, the TL431 is an adjustable zener, but its maximum current is 100mA.

#### ronv

Joined Nov 12, 2008
3,770
Anything short of the switching power supply has to disappate 15 volts at 8 amps -- lots of whatts and very inefficent.

#### GopherT

Joined Nov 23, 2012
8,012
The point is, use a 15V zener for reference (you can connect at 47k resistor in series with it. Since you now have a Vcc - 15V reference, then you can always use a switcher to set the 65VDC supply to 50 or less.