# How to have two separate supply with fixed difference of 100mV??? Need Help...

#### vitz_12

Joined Oct 5, 2011
8
How can I source two different voltage level maximum of 60V with a fixed difference of 100mV between them, coming from a single source capable of upto 90V max. Please see drawing attached.
Do anyone have sample and simple circuit capable of this?

#### Attachments

• 9.7 KB Views: 40

#### russ_hensel

Joined Jan 11, 2009
825
This seems like an odd requirement ( some information on the current desired would also be useful ). Can you say why you want to do this? Note to get 60 out you need to regulate that side as well, or are you using the adjustbiility of the 90 volt supply for that?

#### SgtWookie

Joined Jul 17, 2007
22,220
You need to clarify your requirements.

Does the 0v-90v supply actually have to go from 0v to 90v? Or, can it start from a higher voltage, say 2v or more?

What is the current requirement for output2 (Vsource-100mV)?

#### ErnieM

Joined Apr 24, 2011
8,082
Sounds like a job for two linear regulators with that offset built into their reference voltages (if adjustable) or feedback resistors (if fixed).

The latter is cake and pie. The former would take a few brain cells.

#### MrChips

Joined Oct 2, 2009
22,521
As Wookie says, what is the current requirement?
For example, you can build a 1mA constant current source feeding into a 100 ohms
or 10mA into 10 ohms
or 100mA into 1 ohm.
All will give a 100mV drop.

Last edited:

#### John1

Joined Jul 22, 2011
9
-Simple ?

well you wont get much simpler than this.
I assume the current isn't much.
I assume its DC not AC.
You would have to try a few diodes till you get a drop you're Ok with.

You should give more details,
we shouldn't have to guess or make assumptions.

John : )

#### Attachments

• 29.5 KB Views: 35

#### MrChips

Joined Oct 2, 2009
22,521
Pretty soon Wookie will correct you on this. You cannot get a 100mV drop from a diode unless you come up with some exotic semiconductor materials with very low band gap.

#### ErnieM

Joined Apr 24, 2011
8,082
I've read you can tune a flux capacitor to a precise 100 mV drop. However, you still need the 1.21 gigiwatt driver.

#### vitz_12

Joined Oct 5, 2011
8
I have DUT with differential input pins capable of reading the difference thru ADC down to 2mV - 100mV. My circuit should be capable of forcing 5V to 60V at any pin from the diff pins.. I have external source capable to supply 0 - 90V, in case we need to supply regulator with voltage requirement greater than 60V.
The current requirement of the DUT is 0.3mA for each diff pins.. Don't worry on the current capability of the external source as it can supply enough current to our circuit....

#### vitz_12

Joined Oct 5, 2011
8
I attached drawing if my explanation is not clear...

#### Attachments

• 29.4 KB Views: 30

#### vitz_12

Joined Oct 5, 2011
8
I have available more than one external 0-30V Vsource separate from 90V Vsource. This 30V Vsource is independent to each other. This is just in case we need other external supply needed for the circuit.

If I sweep voltage 5V to 60V at +Vin, the -Vin should be sweeping as well with fixed 100mV difference...

#### ErnieM

Joined Apr 24, 2011
8,082
How well defined is this .3mA load? If it is constant you could just put a resistor of R = .1/.0003 = 333 ohms.

If it's not a constant then you'd be best served with some sort of active divider driving a voltage source. Something like a constant current sink from the 5-60V to it's ground to drive a fixed reference.

#### vitz_12

Joined Oct 5, 2011
8
How well defined is this .3mA load? If it is constant you could just put a resistor of R = .1/.0003 = 333 ohms.

If it's not a constant then you'd be best served with some sort of active divider driving a voltage source. Something like a constant current sink from the 5-60V to it's ground to drive a fixed reference.

thanks for this ErnieM.
the current would be variable and the 0.3mA is max value.

One question though, cause it's not clear to me if this circuit will give a 100mV fixed difference between the +Vin and -Vin, in whatever voltage level I forced on 5-60V source? Please explain further, sorry for my naive question..

Joined Dec 26, 2010
2,148
A 100mV reduction is not easy to achieve with a single device, but what about using two slightly different devices, one for each channel, dropping say 1.25V for one side, 1.35V for the other. This method could have the advantage that the two channels may be more equal in terms of impedance than if a widget was applied to one side only

It is easy to obtain bandgap-based devices like the old LM185 which can be set up to provide accurate voltage drops from about 1.2V upwards, using a pair of resistors. The LM185/285/385 are obsolescent devices, but probably still available easily enough. More modern equivalents should be available*. http://www.national.com/ds/LM/LM185-ADJ.pdf

If more than 20mA is required, this can be augmented with a power transistor (See page 7 of attached data-sheet). Note that the minimum regulated voltage rises to 1.24V +VBE. Currents of more than a few hundred mA would warrant a Darlington (higher VBE). The minimum dummy load current will be also be a bit more: depending on exact volts to say 50μA + VBE/R3.

Edit: *Like for instance the LM4041. http://www.ti.com/lit/ds/symlink/lm4041c.pdf Since I see there is only 300μA max current required, a transistor will not be necessary.

Last edited:

#### vitz_12

Joined Oct 5, 2011
8
It is easy to obtain bandgap-based devices like the old LM185 which can be set up to provide accurate voltage drops from about 1.2V upwards, using a pair of resistors. The LM185/285/385 are obsolescent devices, but probably still available easily enough. More modern equivalents should be available*. http://www.national.com/ds/LM/LM185-ADJ.pdf

Hi Adjuster, using two circuit with 100mV offset between them is good. But do you know something like LM185 that can handle upto 60V? Because I need to sweep the voltage to +Vin/-Vin from 5V to 60V.

Joined Dec 26, 2010
2,148
Hi Adjuster, using two circuit with 100mV offset between them is good. But do you know something like LM185 that can handle upto 60V? Because I need to sweep the voltage to +Vin/-Vin from 5V to 60V.
You would not expect the LM185 etc. to handle more than a few volts. It would be terminated in a dummy load of say 100kΩ to supply a minimum 50μA at 5V. This dummy load current would rise to 600μA at 60V: if you don't want that much drain you could use a constant current load, at the expense of more complexity.

Where you might get real problems is if you have to deal with short-circuits, or even sudden start-ups if there is much load capacitance. That would require the use of pretty clever surge protection. This is harder given the fact that you cannot add much series resistance in the output path if you do not want to upset your precious 100mV offset, although some of that impedance might go inside the reference device loop.

Edit: I see you have expressed a doubt about ErnieM's suggestion, which follows a broadly similar principle: have you understood that all these things seek to create constant series drops, riding below the much larger supply voltage?

Last edited: