Using LM317S adjustable linear regulator for Spartan 3?

There are low cost LDOs that are much better than the LM317. Most notably less voltage drop. Keep in mind that you need to get rid of all the heat that comes with a large voltage drop.

Also note that you can use an LDO with 1.25V reference voltage (for example the lm1086) without any ressitors to power a spartan-3.

Kolja Sulimma

Eric Smith wrote:

Since the input to the 1.2V regulator is going to be 5.0V, I don't really need an LDO. I prefer to use a non-LDO regulator when possible, but in this case maybe I'll really need the LDO because I'm concerned about 1.2V being exactly the minimum specified output voltage for the LM317S.

I wanted to use a regulator available in a TO263 package (surface mount equivalent of TO220) for good heat-sinking. The only 1.2V fixed linear regulator in that (or any similar) package I found was the very expensive LP3881ES-1.2.

Thanks for calling that one to my attention, but I don't think I can use it. The tolerance and the line and load regulation performance could easily result in exceeeding the 1.26V maximium operating spec for Vccint.

The LM1086-ADJ adjustable version regulates to keep the feedback at

1.25V, so that wouldn't be any better.

Thanks! Eric

OK, I assumed that you can influence the input voltage of the regulator.

I prefer to use a non-LDO regulator when possible,

That should not be a problem. You do not even need the resistors.

If you generate 3.3V for your I/Os you can generate the 1.2V from the

3.3V to spread the among to packages. You can also put one or two high current diodes in series with the ldo to burn heat.

For that price you should use a switching regulator. No heat problems there.

Actually that's what I ment. According to Austin you will only experience a minar degradation in device lifetime if you exceed the spec by a few tens of mVs.

Kolja Sulimma

Kolja Sulimma wrote:

True. I was just try> Actually that's what I ment. According to Austin you will only

On closer study of the LM1086 LDO you recommended, I see that there isn't a 1.2V fixed version, and you were most likely suggesting using the adjustable version as a fixed. The spec range of the reference voltage is 1.225 to 1.270V. Taking 0.2% line and 0.4% load regulation into account, the expeced output range is 1.218 to 1.278V. That's below the absolute max rating of 1.32V, so I'm not too worried about damage to the part, but I'd definitely prefer to stay within the recommended operating conditions of 1.140 to 1.260V since the part isn't guaranteed to meet the other specs outside that range.

If I was building a one-off prototype, I wouldn't worry about it.

As it turns out, the LM317S won't meet the spec either. The first page of the data sheet says that it is suitable for output voltages down to

1.2V, but the actual reference voltage spec has a range of 1.20 to 1.30V with a typical of 1.25V.

I'll either use the LP3881ES-1.2 or a switcher, possibly the Linear Tech LTC3406BES5-1.2.

The specs of the LP3881 are quite impressive; I suppose I can see why National is so proud of it.

Thanks again for the advice.

First, you seem to be comparing LP3881 in qty.1 with LM317 in Qty.100. Alternatively, you can take a look at Sharp regulators PQ012FZ series. Digi-Key sells them for $.66 in qty.100

- aap

- The "A" variants of the LM317 have better reference specs ( 1.262 V max at Tj = 25C, 1.270 V over -40 to +125 C )

Which is within an IR drop of Spartan3 VCCINT max for amp-ish load currents, plus both the load regulation and Vref vs temp curves for the LM317 show a decreasing Vout for Tj > 25 C

- IIRC, grounding ADJ on a 317 sets Vout = Vref with no resistors

- also, see the very latest S3 errata for brand new VCCINT ramp rate and supply sequencing requirements

- Nati

After reading that post, I emailed you the offer of a box of XACT 5.x disks and manuals for the cost of shipping, to which you agreed.

The UPS package was delivered to you on April 23.

Some 8+ months later, I'm still waiting for payment or a response to any of my subsequent e-mails.

Brian

You're right, I goofed. I meant to use the Digikey qty. 100 price of $2.581 rather than the qty. 1 price. Still, that's a big difference.

Thanks for the tip. That's definitely an attractive price. I was hoping for a single stage 5V to 1.2V regulator, but the max Vin of the Sharp is 3.7V. Still, I can afford to use a higher current rating on my

3.3V regulator and do it in two stages.

Eric

Thanks for bringing those facts to my attention. Looks like the LM317A may be viable. Though the Sharp regulators Alexi mentioned are quite attractive too. I like the SC-63 package dimensions better than the SOT263, though either will be OK for this design.

My apologies, I haven't seen the email, though I did receive the package. I will find your postal address in my email records and mail you the money tomorrow.

"Eric Smith" schrieb im Newsbeitrag news: snipped-for-privacy@ruckus.brouhaha.com...

with

supplies,

I don't like LDO's and try to avoid them for several reasons:

price, performance, availability, standardization.

My approach is to build a "custom" LDO with a OPAMP (TLC272), a SI2302 (Power Fet), voltage divider to set the output to the desired voltage, a resistor and capacitor for compensation. A reference voltage is already there (from a TPS5120 DC/DC controller). It's very cheap, but needs a few more components. It gives good transient response for the 1.2V Spartan supply, and good analog performance for the 2.5V supply to the DCM's in my case. But going with a LM317 should be good also, not demanding so much "analog" design.

Raymund Hofmann

Have you looked at the ON Semiconductor offerings ? They have quite a few regulators, at very good prices. I use the LM1117 Series quite often. It does not have a fixed 1.2V version, but it does have an adjustable one as well.

Regards Anton Erasmus

I wrote, after being frustrated at not finding any cheap fixed 1.2V: linear regulators:

[I later discovered that it w> Have you looked at the ON Semiconductor offerings ? They have quite a

The National LM1117 and the On equivalent, the NCP1117, have the same problem with reference voltage. For On's part, the max is 1.27V, which is too high because the Vccint maximum is 1.26V.

Admittedly this isn't out of range by much, and would probably be OK, but I prefer to stick to proper engineering practice and use parts with which I can guarantee that the specs are met.

The only obvious problem with the Sharp part is that the data sheet doesn't specify the allowable characteristics of the output capacitor (minimum/maximum capacitance and ESR). For an LDO that seems fairly important.

Eric

Recently someone posted mentioning a Spartan 3 development board available from Altium (for their Protel/Nexar stuff).

You can download a full design package for this board and interestingly they use the same LDO adjustable regulator for all supplies. I can't remember the exact part, an LM1048 or LT1048? something like that.

For Vccint they pull the regulator ground pin slightly below ground by tapping down the -ve rail from a MAX232 type driver on the board.