[time-nuts] DIY FE-5680A lobotomy (disable temp compensation)

[Bob Camp]

Hi

Most of these lightweight Rb’s do the same thing. They watch the oven current on one or the other section and try to guess the external temperature. Based on that guess they do a simple temperature correction on the unit. The older analog units feed a DC signal into the EFC. The newer digital units feed a signal into the DDS.

In both cases (analog and digital), the ADEV of the units can be improved by disabling this “feature”. That of course assumes you are at a constant (as in very constant) abient temperature. In the case of the analog part, it’s the noise on the heater current that gets you. In the case of the digital approach, it’s the tuning granularity of the DDS that messes things up (and possibly heater current noise as well). 

How constant is “very constant”? That depends on the Rb you have. A good bet is that your device runs better than 2 to 4 ppb over a 100C range without the compensation turned on. That gives you 20 to 40 ppt per degree C. To hit 1 ppt you would need to control the device to better than 0.05 C. If you simply want to hit the 0.1 ppb temperature spec, then you only need a two degree control. If you look at the temperature compensation data words (ddd steps), some Rb’s in a batch are much better than others, so there is no easy way to be sure of the results ahead of time. 

Bob
 
On Jun 28, 2014, at 3:21 AM, Magnus Danielson <magnus at rubidium.dyndns.org> wrote:

> Hi,
> 
> I fail to see what the benefit is of removing this unless a better temp compensation scheme is used. It is not likely to interfer with the external loop as it reduces the midterm noise that is systematic. It does add some higher rate noise but that is quantization errors of the systematics it reduces. I like to see measurement that support the claim and I am skeptic. As I see it you give the external loop more systematic noise to dampen and the tighter loop you make the more you will expose.
> 
> Cheers, 
> Magnus
> 
> <div>-------- Originalmeddelande --------</div><div>Från: Scott Newell <newell+timenuts at n5tnl.com> </div><div>Datum:28-06-2014  03:50  (GMT+01:00) </div><div>Till: time-nuts at febo.com </div><div>Rubrik: [time-nuts] DIY FE-5680A lobotomy (disable temp compensation) </div><div>
> </div>Bert asked me to send an update on the FE-5680 tempco mod progress.
> 
> It appears that the FE-5680A temperature signal (or maybe it's really 
> a current sense signal?) can be disabled by removing a single 10k 
> 0805 surface mount resistor.
> 
> Using Elio Corbolante's terrific high-res scans, I've noted the 
> resistor location: http://www.n5tnl.com/time/fe-5680a/lobotomy.png
> 
> Why would you want to disable temperature compensation? As we've 
> seen, the unit's firmware will adjust the DDS frequency as the 
> temperature signal changes. If you're using the '5680 inside a 
> control loop, it's likely to conflict. By removing the resistor, that 
> channel of the 12 bit ADC will be tied to ground through an existing 
> 2.21k resistor. The unit will see a constant 0 counts from the ADC 
> and assume it's really cold.
> 
> I modified one unit and monitored it for a few hours over a range of 
> temps, running it nice and hot with no heatsink, then blasting it 
> with a fan and placing it on an ice-cold heatsink. I observed no 
> change in the DDS tuning words.
> 
> It's a really easy mod--remove four screws, set aside the insulator 
> sheet, and apply your hot leucotome/soldering iron.
> 
> 
> I've also found a simple mod to replace the temperature signal with 
> the output of the unused trimpot. This allows you to simulate any 
> temperature you want. If there's any interest, I'll set up a test and 
> monitor the DDS tuning words as the unit's firmware tries to adjust 
> to the fake temp signal.
> 
> 
> -- 
> newell  N5TNL
> 
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.