In this post I will present the Robocold beta setup a.k.a. the LGAD long term test setup, LGAD stability test setup, etc. The first presentation of this setup I made it about one year ago in this link. After that, other projects appeared, also the climate chamber was failing, so I ended up delaying the development of this setup a bit. But it is finally here!
The goal of this setup is to host up to 8 DUTs (LGADs) simultaneously, all mounted in readout boards and biased all the time at operating voltage. While in these conditions, periodically perform beta scans on each DUT to study their behavior, mainly charge and time resolution, and monitor at all the time the bias voltage and current of each. All this without having to open the climate chamber, so they can be kept cold at low temperatures and operating continuously during a long period of time (several months). Below there is a picture of the setup:
The setup is composed by
- Robocold: 2 axes motorized stages holding the boards with the DUTs in fixed positions and moving the beta source and the MCP-PMT to any position. It has a controller to connect it to a PC and control it remotely.
- The Castle: This is the 8 channel RF multiplexer. 8 inputs, one for each DUT, 1 output to the oscilloscope. It is connected via USB to the computer so it can be controlled remotely.
- Two CAEN high voltage power supplies which are controlled by the computer and implement 8 independent high voltage bias channels.
- A 90Sr beta source.
- A Photonis MCP-PMT which is used as a trigger below the DUTs.
- A LeCroy WaveRunner oscilloscope to digitize the signals.
- A climate chamber to go to low temperatures for tests on irradiated detectors.
- A computer that is in charge of controlling the whole setup automatically.
A simplified block diagram of the setup is shown below.
The beta source and the PMT are mounted such that they move together, as can be seen in the video below:


As of now, the setup is fully operational. I am running it since two weeks in “test mode”, i.e. doing a series of tests to validate it. Also refining the software in charge of the automatic control and data processing.
First testing results
The plots below show some measurements that were taken with the setup during the last weeks. As this post is about the setup and not the devices, I will not pay much attention to them. Only to say that all the devices named LTT__
are irradiated while the device MS33
is not irradiated.
The first plot shows the bias current and voltage of each device as a function of time:
This data was taken during a period of 2.5 weeks. Both the voltage and the current are measured periodically in a time interval that can be configured individually for each device (as some are much more stable than others). The setup measures both the mean value and the fluctuations of both the voltage and the current each time. The plot contains error bars for each point, which can be better appreciated by zooming in. The big fluctuations that can be seen here e.g. in the beginning or on August 17 as well are because I was doing some other tests. E.g. on August 17 the craziness is because I measured a set of IV curves, which are shown below.
As seen, the setup is capable of measuring the IV curve of irradiated devices (all the LTT__
ones) as they have a bigger current, and it can also measure the breakdown voltage of a non irradiated device (the MS33
).
During this period of time a number of beta scans were performed also, but only on the MS33
device which is non irradiated and was also characterized in the past. The plot below shows the collected charge measured for each of these scans. Here we see that the results are consistent between themselves and also persistent along the time, as the first beta scan (blue) was taken on August 16 and the last one (light blue) on August 21.
For the aforementioned beta scans, the time resolution of the DUT (i.e. MS33
) was computed, and is shown below. Here we see that the DUT achieves a time resolution of ~(32±2) ps, which seems to be consistent with previous results. For the time resolution of the reference_trigger
, which is the MPC-PMT, a value of 17±2 ps was considered. This number comes from a previous measurement I performed on it.
Further development
As the setup is ready to start working, the only remaining part is the readout board. I designed the Chubut board an year ago for this, which is actually the board in use right now. However, soon I discovered that the available space to fit the samples is very small (I had to do some dirty tricks for the Market Survey samples, and also other samples are not fitting now). So recently I made a version of such board[1]Version 220815 found here. which has a lot more space for the samples.
As the new version of the board is still single channel, and for many applications it would be nice to have multi channel boards, I am also developing the Chubut 2 board which should be similar but with two channels. I also want to include a few other features that will make the life of that guy who is in the lab (i.e. me) much easier, the main one is to aim to a design with a carrier board to mount the samples which in RF is far from trivial.
