Buffer for a DAC 5

[jamesnguyen]

Hi,
I am using a DAC to control the voltage applied to "Iset" of a white LED driver IC (Onsemi CAT4104). Unfortunately, the DAC does not have enough current driving capability the CAT4104 requires (about 1mA).

I think I will need a buffer for the DAC. An Op-Amp buffer will do the job, but I am wondering if there is any cheaper, simpler transistor-based implementation out there?

Regards,
James

 

[IRstuff]

Define "cheaper"

What kind of DAC is it?

TTFN

FAQ731-376

 

[Skogsgurra]

What DAC cannot supply 1 mA? An R2R? Or a PWM with filter?

The cheapest buffer would be an emitter follower (one transistor), but you will not get it very accurate and you must compensate for the around 700 mV base-emitter voltage drop, which also is temperature dependent and non-linear.

Next step is a unity gain opamp. One component again, but a little costlier. Good precision can be had.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[itsmoked]

You can find SOT23-5 OPAMPS that take little space and shouldn't cost much at all.

45 cents in onezzzz

20mA

http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=MCP6L1T-E/OTCT-ND

Keith Cress
kcress -

http://www.flaminsystems.com

 

[benta]

The CAT4104 does not have an "Iset" input, but an Rset, which has a fixed voltage.
Sounds to me like you are using this device _way_ outside datasheet specs.
Instead of your DAC, a digital potmeter would be more to the point.

Benta.

 

[jamesnguyen]

Hi,

IRStuff,
Cheaper means few transistors instead of an op amp.
It's a Aux current DAC from the AD6855 CODEC.

Benta,
Yes, I meant "Rset". The device's LED current is controlled by adjusting a resistor value from "Rset" to Ground. I don't see why I can't connect disconnect this resistor from ground and connect a DAC to it. Do you see a problem?

 

[IRstuff]

But why is that cheaper? It's not just a "few" transistors, it's also resistors and capacitors, all of which need to be picked and placed, even with an automated assembly system. Moreover, you'd drastically increase the inventory control from a single component to around 5 components.

As for Rset, did you read your datasheet carefully? It implies that the Rset pin is tied to a voltage reference, which means that you cannot program it with a voltage. This is reaffirmed in Fig. 4 of the datasheet.

TTFN

FAQ731-376

 

[jamesnguyen]

I think I have problems explaining it clearly. The picture below shows my intended use.

 

[itsmoked]

You're not getting it James... You can not control the LED brightness by using a varying voltage into that pin.

Let me repeat: [!]YOU CAN NOT CONTROL THE LED CURRENT WITH A VOLTAGE TO PIN "RSET"[/!]


Throw the DAC out entirely and replace it with a 10K ohm digital potentiometer as benta wisely suggested. They are easier to control anyway.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[jamesnguyen]

OK. Let me try one more time.

The DAC is NOT connected to the "Rset" pin directly - it connects to it through a resistor (says 10k).

Now, if the DAC output is 0V, it would be the same as if the resistor is connected to ground, isn't it? If the DAC voltage increases, I think it would appear to the LED driver as if the resistance is increased.

Maybe you are right, but I will do the test anyway just to be 100% sure. Thanks!

James

 

[IRstuff]

Is this for production? Your ability to control the low-end LED current is problematic, since the Rset pin voltage is not guaranteed to be accurate.

TTFN

FAQ731-376

 

[MacGyverS2000]

Why would you ever control LEDs this way...?

Dan - Owner

http://www.Hi-TecDesigns.com

 

[itsmoked]

Not sure what you mean Dan, I do it all the time.

A lot of controllers; LED, V regulators, have that damnable ground resistor current setting.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[benta]

So many strange things about this design.
10 kohms is right at the datasheet limit.
If there is a D/A, there is probably a controller. The CAT4104 supports PWM dimming: free of charge.
10 kohms and 1.2 V Rset voltage can never make 1 mA.

And so on.

I give up.

Benta.

 

[MacGyverS2000]

Keith,

benta is thinking like me... this design appears to be going around your elbow to get to your back-end, with little to no perceivable benefit going this route. For things like this, I like to know exactly what the end application is, and quite often it turns out there is a more simple/cheaper way of doing it...

Dan - Owner

http://www.Hi-TecDesigns.com

 

[IRstuff]

The correct, datasheet resistor values should range from 10 kohm to get an LED current of about 20 mA, down to 686 ohm to get an LED current of about 175 mA.

If you were to ASSUME the nominal 1.2V R.set voltage, you could get the same operating range with a 680 ohm resistor, and the free of the resistor at a range of voltages from 0 V up to about 1.064 V.

However, there is a variance on the Rset voltage, so any open loop control will have variance in LED output at the low end.


TTFN

FAQ731-376

 

[IRstuff]

Alternately, you start with a digitally programmable LED driver, ala

http://www.allegromicro.com/en/Products/Part_Numbers/6281/6281.pdf

I'm sure there are tons of others, as well.

TTFN

FAQ731-376

 

[jamesnguyen]

Thank you everyone for your responses. I really do appreciate it.

Here is the story:

On one of our existing products, the LED driver (Exar SP7616) is going to be EOL. This driver has the current gain of about 1000. In other words,

ILED = (1000 * 1.2)/RSET

We are currently controlling the brightness by connecting a DAC to the other end of Rset and this method works well so far.

The replacement candidate is the CAT4104. Unfortunately, this driver has a current gain of 100 or,

ILED = (100 * 1.2)/RSET

Because our DAC has output impedance of 7.5k, we can't really get the required 20mA since the LED driver needs 6.34k to ground to produce 20mA.

That's why I was asking for a DAC buffer at the first place.

Other points:
- Dimming using a digital pot is also not possible because there is no "free" serial port available. We could run an I2C bus there, but we want to minimize the changes on the existing design.

- PWM dimming is not "allowed" by the RF group because it could cause desense issue. The radio needs to work down to -140 dBm (well, that's what I am told)

- According to the datasheet, Vrset varies from 1.17V to 1.23V. I think I can work with that.

- Don't give up yet.

- No, this is not a homework assignment!

 

[itsmoked]

Dimming using a digital pot is also not possible because there is no "free" serial port available. We could run an I2C bus there, but we want to minimize the changes on the existing design.

Most digital pots don't need serial ports, or I2C, or SPI.
Just two wires. Direction and a number of pulses in the direction of interest. Scrapping the expense of the DAC and the wires needed to run the DAC should make available the simple resources required to run the aforementioned digital pot. It would also reduce the foot print. It would require a minor firmware change...


Interesting problem you have.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[DHambley]

Itsmoked has the best suggestion, a digital potentiometer. Many DACs are exactly just that. The output is simply a resistance to ground and the voltage is created by a current source..you just leave the current source portion out.