Testing of NMR and MRI Console Electronics
(Nuclear
Magnetic Resonance and Magnetic Resonance Imaging)
Robert Gilchrist Huenemann, M.S.E.E.
(831) 635-0786
January 26, 2006
Between 1971 and 1973, I was the principal engineer for the Hewlett-Packard 9540 Transceiver Test System. The 9540 used the then new HP 8660 RF signal generator, and HP was having some difficulty manufacturing this unit. Since the 9540 also included an HP 432 power meter and an HP 5326 frequency counter, I found it advantageous to write an automated test for the 8660, and eventually for other parts of the 9540 system as well. This allowed me to quickly return defective instruments to the various HP manufacturing divisions, with a detailed listing of their failures.
Between 1973 and 1980, I was employed in avionics maintenance support at United Airlines. While there, I developed an automated system for recovering VOR (very high frequency omnirange) bearing angles from the Fourier transform of a digitized VOR signal.
These two experiences came together in 1990 when I interviewed for a manufacturing test engineering position at Varian NMRI. It was immediately apparent that NMR and MRI console (and magnet leg) electronics are ideally suited to an automated selftest process. I suggested the development of such a test, and was hired to do this development.
These tests are now in use on the Varian Inova, Mercury and InfinityPlus product lines, and they have been developed to the point that they incorporate thousands of test steps which verify the proper operation of every part of these products. They are the only system level tests used by Varian in the manufacture of these products, other than a short initial diagnostic sequence which verifies that the host computer can communicate with the acquisition computer. No NMR tests are done on Varian NMR consoles until they reach the customer, and the many NMR magnets which were formerly used by Varian for console test have been eliminated, for a cost savings of several millions of dollars.
The selftest verifies every gain and level step, every phase angle down to 0.25 degrees, all modulation waveforms, and a select group of frequencies chosen to verify each of the frequency programming bits. It verifies all shim and PFG currents, and all waveform generator outputs, using a small data acquisition system. The only other test equipment required is a set of low power RF power dividers and attenuators.
The test is semi-automatic in that it requires an operator to make various system connections, and to insert appropriate values of attenuators into signal paths. However, approximately half of the test steps are made with one single set of connections that includes every system signal path, such that this part of the test can be run repeatedly until interrupted. This portion of the test is typically run for 24 hours or more and has proven invaluable in identifying intermittent failures, which are of course the most expensive failures to remedy in the field. Such failures have essentially been eliminated from the Varian product line.
Many of the required test steps were apparent to me, based on my previous experience. However, I also have read every Varian installation report for the past decade. Occasionally, a failure at installation has identified additional test possibilities. This and the introduction of various features such as multiple receivers and coordinate rotation has led to the present exhaustive sequence of thousands of individual test steps.
The tests include a simple user interface that allows the operator to stop on failures, to repeat tests for troubleshooting purposes and to enter information for the system records. These records include a summary of system failures and a complete record of all system tests.
Finally, one small but crucial detail: To implement an automated selftest of an NMR system, it is necessary to turn on both the transmitters and receivers simultaneously. I started the Varian project with the Gemini product line, which included hardware gates that turned off the observe transmitter during an acquisition. They could only be circumvented by rerouting soldered connections. The decouplers gates were not hard wired to the receiver, so I was able to write extensive tests for the decouplers. While this allowed me to convincingly demonstrate the concept, it was not practical to do this unsoldering and resoldering in production.
As a direct result of this demonstration on the Gemini, the Varian Inova, Mercury and InfinityPlus product lines were developed with gates that allow all transmitters and receivers to be turned on simultaneously, including the observe and lock channels. Every one of these individual gates is accessible for test purposes. Of course, this level of gate control is available only by means of special instructions that are not provided to NMR customers.
NMR and MRI Selftest
– Some Details
I can help you to make your manufacturing tests more comprehensive, and I would appreciate an opportunity to discuss this matter at your convenience. I would also appreciate it if you would forward this information to anyone else who might find it to be of interest. You can reach me at the e-mail link on my home page.