LTZ1000 Supply

[e61_phil]

Hi,

I'm working on a LTZ1000 reference and I would like to transport this thing to an calibrated 3458A without powering the reference down. So, I need a battery solution. I think three sealed-lead-acid 6V batteries could work fine. One can also use the batteries for precision measurements without any ground loops for days.

The build in power supply should keep the batteries (current limited) on 20.4V (standby use of the batteries) if connected to line voltage.

My question is: should I supply the LT1013 directly with the approximately 20.4V from the batteries and rely on the PSRR of the LT1013 or should I insert a LM7815 to isolate the changing battery voltage from the LT1013?

Thx
Philipp

[Andreas]

Hello,

see my measurements here (and the following post)

https://www.eevblog.com/forum/metrology/ultra-precision-reference-ltz1000/msg842662/#msg842662
I have around -0.3 .. -0.4 ppm/V on my samples without stabilisation.

With stabilisation I have a positive PSRR 0.1 .. 0.2 ppm due to the heating of the SMD voltage regulator on the pcb.
https://www.eevblog.com/forum/metrology/ultra-precision-reference-ltz1000/msg846835/#msg846835

So you should keep power consumption as stable as possible near the reference.
(so the LM7815 far away from the LTZ).

With best regards

Andreas

[e61_phil]

Ok thanks

I will integrate a voltage regulator out of the LTZ1000 housing.

[acbern]

If you transport it, you better transport it to a calibrated voltage reference, this gives you much better accuracy than a 3458A. There are certainly some guys here, depending on where you live exactly in G, who would support you with this, including myself.

[e61_phil]

The calibrated HP 3458A is the best I can get here. But your suggestion sounds very good.

I'm from Bremen. Is there anybody around here?

[uncle_bob]

Hi

If you are going to all this trouble, toss in a temperature sensor of some sort. Why? Well, you drove to someplace with it in the back of the car. You carry it a ways (at -10C) to get it into the new location. Now the question: has it stabilized to room temperature (yet) or not? It is not just a matter of temperature stability. Hysteresis and gradients are an issue as well. You want to wait a bit *after* it is stable before you do the calibration stuff.

Bob

[e61_phil]

I planed to ovenize the LTZ1000 compartment for that reason. I thought about 30°C to be not to close to the 45°C setting of the LTZ1000. The other reason are wirewound resistors with a tempco of 5ppm/K instead of Z-foil resistors.

[uncle_bob]

I planed to ovenize the LTZ1000 compartment for that reason. I thought about 30°C to be not to close to the 45°C setting of the LTZ1000. The other reason are wirewound resistors with a tempco of 5ppm/K instead of Z-foil resistors.

Hi

Even so, there will be part of the gizmo outside in the room. A cheap thermistor to attach to a DVM is a really simple thing to do. Just about any DVM will be accurate enough to read it out closer than you would ever need.

This sounds nutty, but it is an issue. When you do your 40C oven, run it off of an independent battery pack. Isolate the supply and ground for the oven from the rest of the circuit. That way the variable oven current does not mess up the precision side of things. It is not just static offsets. Oven current tends to be noisy. Also if your 40C oven dies due to a flat battery, your precision side is still stable and running.

Bob

[e61_phil]

Ok, I will place a thermistor inside the device.

When you do your 40C oven, run it off of an independent battery pack. Isolate the supply and ground for the oven from the rest of the circuit. That way the variable oven current does not mess up the precision side of things. It is not just static offsets. Oven current tends to be noisy. Also if your 40C oven dies due to a flat battery, your precision side is still stable and running.

Bob

Do you really mean this is neccessarry, even when I have a separate voltage regulator between the oven and the LTZ1000 supply? I would like to use a complete linear oven control to avoid any swithing noise inside the box.

The flat battery is a good argument, but I think it is possible to switch off the heating before the battery is completely flat and remain enough power for the LTZ1000 circuit.

[uncle_bob]

Ok, I will place a thermistor inside the device.

When you do your 40C oven, run it off of an independent battery pack. Isolate the supply and ground for the oven from the rest of the circuit. That way the variable oven current does not mess up the precision side of things. It is not just static offsets. Oven current tends to be noisy. Also if your 40C oven dies due to a flat battery, your precision side is still stable and running.

Bob

Do you really mean this is neccessarry, even when I have a separate voltage regulator between the oven and the LTZ1000 supply? I would like to use a complete linear oven control to avoid any swithing noise inside the box.

The flat battery is a good argument, but I think it is possible to switch off the heating before the battery is completely flat and remain enough power for the LTZ1000 circuit.

Hi

The problem is not just the supply voltage. It is the current flowing in the ground. A change in current gives you a change in voltage. If the grounds are in common, you get an output bounce. You can go crazy with various wiring techniques to combat this. It's easier to design it with an isolated ground and be done with the problem.  The 40C oven will pull way more power than your reference. It's likely that you will need 10X the battery for it than for the reference. Adding the extra cell is not going to be a very big deal.

You can say -- what if I insulate the outer oven *really* well? Ok, but now the "dead power" (power your control loop can not drop out) will drive up the temperature of the outer oven. You also don't want the outer oven to run away when it's 35C on a hot summer day. Unfortunately there is no magic solution to this.

Not just a theoretical issue. I've got a lot of empirical data on how to do this wrong ....

Bob

[e61_phil]

I would design the circuit in a way that the voltage regulator has the same GND as the reference circuit. The GND for the heating is on a star point before the 15V regulator. So any drop in the ground should be regulated by the 15V regulator.

I would like to use three 6V 4,5Ah batteries, so there should be enough power even for days. I think an extra cell is a big deal in sealed-lead-acid. It is heavy and big.

My problem with separated batteries is the charge circuit. I have to build two isolated charge circuits as well if I would insulated all the things.

I think 35°C on a summer day isn't a big problem in germany

Thank you very much for these tips. I think I will build up the heater with a 15V regulator for the reference and do some measurements on the 15V rail during extreme changes in heating power. After this, I could decide which way I should go.

[uncle_bob]

I would design the circuit in a way that the voltage regulator has the same GND as the reference circuit. The GND for the heating is on a star point before the 15V regulator. So any drop in the ground should be regulated by the 15V regulator.

I would like to use three 6V 4,5Ah batteries, so there should be enough power even for days. I think an extra cell is a big deal in sealed-lead-acid. It is heavy and big.

My problem with separated batteries is the charge circuit. I have to build two isolated charge circuits as well if I would insulated all the things.

I think 35°C on a summer day isn't a big problem in germany

Thank you very much for these tips. I think I will build up the heater with a 15V regulator for the reference and do some measurements on the 15V rail during extreme changes in heating power. After this, I could decide which way I should go.

Hi

I've *been* in Germany when it was 35C on the thermometer in the garden. (No I won't mention what *kind* of garden. I was taking care of a "lack of fluids" problem though.). The bigger issue is in transportation. If the gizmo is in the sun, you have solar gain. It can be a real pain if you have to baby the thing all the time.

Bob

[e61_phil]

The bigger issue is in transportation. If the gizmo is in the sun, you have solar gain. It can be a real pain if you have to baby the thing all the time.

Of course, but the only solution I can think of is to increase the oven temperature. But I would like to stay at 45°C for the LTZ1000. So I think there is no way out?

[uncle_bob]

The bigger issue is in transportation. If the gizmo is in the sun, you have solar gain. It can be a real pain if you have to baby the thing all the time.

Of course, but the only solution I can think of is to increase the oven temperature. But I would like to stay at 45°C for the LTZ1000. So I think there is no way out?

Hi

The solution revolves around insulation.

If you have the LTZ running it's power inside another oven, there will be an un-avoidable temperature rise in that oven. There is also a practical (control loop) limit on how close you can get the outdoors to your oven temperature. As you increase insulation on the outer oven (to reduce power / get more time on battery) you make both the heat rise and control loop problem worse.

With a simple controller and modest dead power, something in the 5 to 7C range is a good idea for control margin. A lot depends on how well you want to control things. The dead power side is a bit easier. If you insulate so that 2W heats you up from -10C (outdoors on the porch right now), that is 25C / W going to 40C. If you go crazy and run 0.2W you have 250C / W. Multiply that times the power the LTZ and associated "stuff" is running. Picking 0.1W totally at random. In the first case, it's not an issue. In the second case you can't even operate at room temperature.

Bob

[e61_phil]

Do you think about sending by mail or why should I heat from -10°C? If one would like to send the reference by mail you're completely right. Then it should last for days even at very low temperatures.

But, my thoughts were on a journey by car, to another reference. In an air-conditioned car the problems are much smaller. If neccessary it is also possible to charge the reference in the car with a switching power supply. The switching noise during journey shouldn't be a problem.

I will build up the oven and make some tests to get a feeling for what is going on.

[uncle_bob]

Hi

If I was going to mail this beast, I'd turn off the outer oven. Let the LTZ do it's thing and self heat. Once it got to the destination, turn on the outer oven and let it run for a few days.

Bob

[Kleinstein]

Normally one should get away with just a 12 V(actually 13.xV)  battery - no need for the extra 6 volts. This might need an extra JFET to extend the output range of the OP, but still less effort than a 3rd 6 V battery. You can still stabilize to 12 V with a low drop regulator - outside the oven. Lower voltage also reduces the power dissipation and thus the temperature rise at a give level of insulation.

One could even consider having the LT1013 outside the extra oven - its drift is not that critical. It's more about the resistors and the PCB that might want a stable temperature. The transistor(s) controlling the heater of the LTZ1000 should also be outside, even though in a already heated environment it will not get that hot.

Even if the extra oven looses control at extreme temperature (too hot or less likely too cold), it can still reduce the temperature swing a lot. So I don't think you have to design for 35 C outside temperature.

When taking all the effort to keep the reference hot, you may also want to have a buffer amplifier, as not to disturb the reference by a to heavy load that might upset temperature regulation. The minimum would be limiting the internal heater to rather low power.

[e61_phil]

I measured 10,4V on the heater output of the LT1013 on my stabilzed LTZ1000 Board, so I think 12V is very tight. Especially if one need a little bit more power in an cold environment or during heating up. I thought about low side heater controlling of the LTZ1000 but I didn't dare to change anything in the LTZ1000 circuit.

With 12V only one 12V lead-acid-battery will be needed, much nicer..

I would like to keep everything close together, so I would keep the LT1013 in the same housing as the LTZ1000.

I already use a LTC2057 as output buffer amplifier.

[quarks]

about the battery, I would probably try to use a power tool battery, because I have quite a few, so I can easily charge and change them very fast.

[e61_phil]

about the battery, I would probably try to use a power tool battery, because I have quite a few, so I can easily charge and change them very fast.

Nice idea and an internal power supply to keep all the things running during change?

[uncle_bob]

Hi

This really messes things up, but the further we go, the more relevant it is.

How about a (duck and run) switching power supply?

That frees you up quite a bit on the battery voltage. It takes care of the "local power" switch over issue (use diode voting). You can get some pretty quiet parts from Linear that come in quite small packages. Filter it / decouple it / follow it with a LT1764. You should be able to keep it from making much of a mess.

Bob

[e61_phil]

Hehe, from two separate batteries for heating and reference to a switching power supply....

I will start with the oven and collect some experience

[uncle_bob]

Hi

It's a design, of course you go all over the map digging into things.

Bob

[Andreas]

How about a (duck and run) switching power supply?

That has the good side effect that you can adjust the output voltage just by holding your hand near above the output lines.
(the disadvantage is that you should not breathe to keep the output voltage stable).

With best regards

Andreas

[uncle_bob]

Hi

I would have no expectation of using the switcher directly for anything. I'd have it generate a bulk supply and then regulate that with something I trust. I'd then re-regulate that supply further down the line for this and that need.

That said, I have not seen a major issue with stability on the Linear all in one parts. Since they have a lot of parts, this:

http://www.linear.com/product/LTM8055 is the sort of thing I'm talking about.

Fine for something that is happy at a percent. Nonsense for something that needs 100uV or less. Enough of it is self contained that the "noise loops" are not going to be gigantic. I normally put coils on output and input before I do anything with similar parts.

Bob