Procurement of the chips turned out to be a lot more difficult than we expected. We literally had to run after some of the parts to buy them. We also had to modify the final PCB several times with the replacement parts. We had problems with the following parts:
AVA-183+ by Minicircuits: We were using two of these amps in the output stage. It is currently out-of-stock at Minicircuits. We must really have some bad luck. In all these years of designing RF circuits, this is the first time a part by Minicircuits is out-of-stock :( We replaced these two amps with HMC451LP3E by ADI and MAAM-011101-TR1000 by MACOM. This led to an increase of about 11 USD per board. However, this new amp combo (MAAM + HMC451) is a lot more power efficient than two AVA-183s.
MC12080D by On semiconductor: We are using this part to generate the 10MHz REF OUT signal from internal 100MHz VCXO. It wasn’t available on Digi-key for some time. Later it got back into stock and we bought it. Waiting for it to get back into stock was a little stressful as it is one of the ICs that doesn’t really have a direct replacement, i.e., there are not any other divide-by-10 prescalers on the market.
LMX2594 by TI: This is the part which will delay us about 30 to 45 days. We have to get this from local TI distributor EMPA. It is available on some of the global distributors but the price is almost twice. When we decided to use this part several months ago, this part was in stock at TI’s factory. Now TI has given us a delivery date of Oct 4th. We placed the order for 250 pieces. Meanwhile we ordered about 10 pieces from Digi-key for final prototype run.
RFSA2113 by Qorvo/RFMD: This part used to be available on some global distributors when we did the campaign proto. Now it is only offered by rfmw.com. We had to use an intermediary firm to buy it for us from rfmw.com.
In summary, we are almost done with procuring. We just need to get the LMX2594 ICs. EMPA promised to deliver them around Oct 4th.
After the completion the campaign run, we made a test PCB just to fine tune the 50Ω transmission line. We were not getting a very nice 50Ω in the campaign prototype at higher frequencies (above ~10GHz). We had to do this because generating +15dBm at 15GHz (which we promised in the campaign) is a little bit challenging when our simple transmission line is not nicely matched to 50Ω. So, we placed transmission lines in both microstrip and CPWG form onto a PCB with exactly the same stack-up as ERASynth and varied the width and spacing. When we got the test PCB back results were a little disappointing. They were better than campaign prototype, but still far from what we were aiming to get. So, we had to repeat the test PCB run. At the moment, the final prototype is ready but we are still waiting for our second test PCB to fine tune the 50Ω track width. Here are some photos of our test PCB and 3D image of final prototype PCB:
We got an intern from a local university and we gave him the task of developing a PC GUI for ERASynth. We didn’t promise this during our campaign and were just planning to release the serial command set. We will make the sources of PC GUI available on our Github repo as well. Here are several screenshots of the PC GUI:
We decided to add a microphone interface to ERASynth to be used with modulations. Inputing a sound wave into any signal generator is a frustrating undertaking. We realized this when we were dealing with modulations. We believe adding a microphone interface would be a low-cost but high-impact feature.
Overall, we did everything we could to complete the project on time, but producing two test PCBs and procurement of some parts will delay us about 30 to 45 days. We apologize for this delay.
We will release more frequent updates from now on. Please keep following the campaign page.