Valhalla 2720GS

[branadic]

And a few more...

[tautech]

Maybe the mods are gentle enough to move that thread into the metrology section? It's better located over there.

-branadic-

Not without endorsement from the OP or they can do it themself with the Move Topic feature only they can see at the foot of each page.

[branadic]

Thanks to the mods for moving the thread 

I still hadn't the time to take high-resolution images of all boards, but will do before I assemble everything.
Any updates to the HV-transformer analysis, Kosmic?

I'm still in contact with several companies, but without definitve result so far. I'm also considering having a ring-core transformer being made, but they asked for power consumption of all rails. I've sent them the image with the resistor values of each winding and still hope to get some answer. 

-branadic-

[branadic]

Good news:

I have a company at hand, that is able to analyze and replicate the transformer at reasonable costs. Only drawback is, the transformer will be 15mm longer, as these core sheet length is no longer available, which results in slightly more power being available. A rewinding would be way more expensive, since the coating has to be removed first, which takes a lot of time.
Analyzing the transformer is about 150€ and a copy costs 135€, so 285€ maximum, that's what I was told on the phone.

So if multiple people come together, we could share the costs for analysis and get working copies. Who is in?

-branadic-

[branadic]

So I have placed an order for 3 HV transformers today. Final dimensions are 95 x 8 0x 70 mm³. I will now package the defect one and send it to the company for analysis.

-branadic-

[branadic]

The transformer is at the company, they will start with their work next week.

Question to the German followers, is someone here with a programmer for Xicor X2210, X2212 and willing to lend it to me? The unit is full of those NOVRAMs and I would like to save their content.

-branadic-

[branadic]

I today received a BeeProg+ from a friendly forum member (spent for a few days) and read the 8x NVRAMs distributed in the unit, with the files in the attachement together with an image of the content of reference 1 vs. 2 vs. 3 (from left to right). First two rows are identical.



Meanwhile, a first transformer copy was fabricated and will make its way to me tomorrow for testing whether or not it is working.

-branadic-

[branadic]

In a former post the following image was drawn:



What is missing here are the two wires connected to E9 and E10 on the Mother Board Assembly and the additional wire that is directly connected to the transformer core with a ring lug. maxwell3e10 could you try to find out how they are connected and if they are connected together? That would explain at least something.

Today, a first sample of the HV-transforer arrived, that only has one shield wire, so it is missing the two additional wires for E9 and E10. Need to talk with the manufacturer why. Other than that the transformer looks nicely build.

-branadic-

[branadic]

I had a phone call with the manufacturer... the additional two wires were screens (copper foil) between the secondary windings (32 Vac with CT and 115 Vac as well as 115 Vac and 1100 Vac). In the sample I currently have only the screen between primary and secondary is in place.

In the next samples they will add the screens between the secondaries too, so that the transformer is as original as possible. However, their technican was the opinion, that such additional screens were not necessary, as they are only good for high frequencies.

Another important thing to note is, that they found a palette with the absolut identical core sheets, so the outer dimensions of this replacement are identical to the original transformer, which is great news.

-branadic-

[branadic]

Have checked the transformer yesterday by resistance measurements, but also by voltage measurements and found, that something was odd and different to the schematic I've drawn in my previous post. User maxwell3e10 already confirmed, that all secondary windings should be symmetrical to the center tap, except for the 1100 Vac.
Unfortunately, the 230 Vac CT (2 115 Vac) are not as they supposed to be on the sample they have sent. I already wrote the manufacturer yesterday and had a call with him this morning. They will now implement it hopefully in the correct manner including the screens and will send a new sample by the end of this week, that I hopefully receive by the beginning of next week.

So I'm kind of lucky to have made the sanity check first, before installing the transformer in the instrument --> better safe than sorry

-branadic-

[Kosmic]

It's nice of them to send prototype like that. Looking good so far 

[branadic]

Yes and the bill that came with it is for the first working sample. By the way, a new sample is ready for shipment and will arrive by the beginning of next week. This mornings call revealed, that in theory everything should be as drawn, continuous winding with center tap as well as screens. Fingers crossed.

-branadic-

[branadic]

From what we know we can now update the transformer scheme.

There is a fist screen (white/brown) between primary and secondary 1100 Vac, a second screen (white/green) between 1100 Vac and 32 Vac CT and a third screen (white/yellow) between 32 Vac CT and 230 Vac CT. The first screen is tied to the core of the transformer, while the other two screens are connected to E9 and E10, which themselves are interconnected on the Mother Board Assembly.

-branadic-

[branadic]

Today a new sample arrived and connection-wise it looks about right. Tomorrow I will power it up and measure the voltages.

-branadic-

[branadic]

I today performed the voltage measurements and it all looks fine. I'm already in the process of reassembling the unit, but had to stop at some point today, but will move on in the next days. Fingers crossed we have a working 2720GS soon.

-branadic-

[branadic]

She's alive

-branadic-

[Kosmic]

She's alive

-branadic-

Ho Nice!

Good Job 

[TheSteve]

Show us the volts!

[branadic]

Show us the volts!

Just for you Steve.

-branadic-

[maxwell3e10]

Would be useful to see a recording as a function of time, to see if there are any spikes associated with cross-comparison, as I was seeing, and general level of drift and noise.

[branadic]

I performed some measurements yesterday and compared it to our 3458A, for which I had to estimate new correction coefficients, based on the history I have so far. Unfortunately, I don't have a valid baseline point for this year by now. A deaper expaination about how I did that and the results I've found so far coming soon in a full repair article on xdevs. However, I found that both instruments agree within <2 µV/V for 10 V, which would be pretty good.

A fun project would be to replicate the reference card and equip it with ADR1399KHZ. There is a free slot 4 on the GS and I have copied the content of the NOVRAM to some spare parts I've bought, so I could populate it with the original drift history of another reference. I guess at the beginning this 4th reference would be rejected by the instrument, due to the fact that, compared to its drift history, the instrument would notice a sudden jump for thiis card. But I guess at some point, assuming constant calibration to an external trustworthy reference, it could accept the new reference board. What is the opinion on that?

-branadic-

[branadic]

0.1 - 10 Hz noise measurement @ 10V with all 6x LM399 being active, so about 8 - 9 µVpp.

-branadic-

[branadic]

I made some progress
Reverse-engineering complete, schematic needs some clean-up though before I can post it. So the way of having a 4th reference board is created, but even ADR1399 upgrade is now possible and other sorts of things that come into mind.

-branadic-

[branadic]

This is the best I could arrange with a limited amount of time.

-branadic-

[branadic]

A first reference board copy is almost good to go.

-branadic-