![]() |
Over the past few weeks I’ve had time to examine many dozens of diodes, mostly germanium, in my crystal radio diode collection. Many of them were removed from equipment built in the '50s and '60s (old diode matrix boards), some are vintage NIB 1N34As while others are modern SMD Schottky style diodes.
There are numerous excellent websites such as this one by Dick Kleijer or SV3ORA's site ... all describing elaborate ways to determine which diode is ‘the best one’ (the holy grail diode!) for crystal radio work. Most methods use a vigorous, somewhat complex test procedure plus a lot of math, most of which is well beyond my old brain, in attempts to flesh out each diode’s inherent characteristics ... as the sites referenced above illustrate, the simple appearance of a crystal diode belies its complexity and determining diode behaviours can be more challenging than one might suspect.
My testing procedures were much more basic, and in the end, may hopefully reveal the best diode in my collection. I think one needs to undertake this with the understanding that there really is no overall ‘best' crystal radio diode but rather, only a diode that is best for your particular system and what works best in my system may not necessarily be the best one in yours.
My plan was to measure a few diode behaviors, shrink the list of candidates and then compare them against each other in my system's high-Q tank circuit.
My first step was to measure Vf or the forward voltage needed to ‘turn the diode on’. This can usually be determined to reasonable accuracy by using the diode test function on most digital multimeters. I’ve always supposed that the diode with the lowest Vf turn-on threshold would probably be the most sensitive, but is it the only factor? Hopefully my tests would indicate if anything else is in play.
The next task was to determine the minimum signal level of a 1000 Hz modulated carrier on 1400 kHz that could be detected by each candidate diode. An RF probe was used to measure the level of signal capacitively coupled into my crystal radio’s antenna tuning stage which was then lightly coupled into the detector stage, using the diode under test. No importance was given to the actual base level of this signal other than to note the level at which it could first be detected by ear (using sound powered phones) and making sure the coupling distance between stages remained the same for all diodes under test. This allowed me to compare weak-signal diode ‘sensitivity’ to the diode’s previously measured turn-on point or Vf value. Would the diode with the lowest Vf also be the most sensitive when used in a detector circuit composed of complex impedance, resistance, reactance and capacitance values that the test diode would be looking into?
The RF signal coupling was adjusted so the injected carrier could be varied between 0 and 10mV as measured on the RF probe. For each diode, the signal level was slowly increased from ‘0’ until the 1400kHz tone-modulated AM signal could first be detected.
The lowest 'first detected' signal level was .6mV while the highest level required 3.4mV, representing a pretty good range of diode behaviours. There were 49 different diodes in the test pool.
Four of the 49 diodes detected the .6mV signal, six detected the signal at .7mV, and nine first detected the signal at .8mV. The remainder required a still higher level of injected signal. The average level of first detection was 1.2 mV.
Of the four .6mV ‘best detectors’, their turn-on Vf values ranged from .15V to .38V while the .7mV and .8mV detectors had a Vf between .181V and .40V!
It seemed, not surprisingly, that generally the higher the Vf turn-on threshold, the greater was the level of signal injection needed for first detection … but evidently using the Vf value alone to determine the ‘best diode’ was not the hard axiom I had always assumed it to be!
Since a low Vf was not necessarily needed for good sensitivity, would there by any other tests that might indicate best performance?
The next trial was to measure actual diode currents in my hi-Q detector while receiving a lightly-coupled constant level input signal (1400kHz) to see how this value related to Vf. Measured diode currents (Id) varied from 9uA to 14uA for the same level of input signal, with the diode having the lowest Vf also producing the lowest current level ... hhhm! There was more to this than I expected, but generally, the lower valued Vf diodes tended to produce the most current and consequently the louder headphone signal … but not always! Some diodes with a Vf as high as .46V yielded high currents!
This now begged the question, “Does the higher current diode with a higher turn on (Vf) prove to be a better overall performer than the diode that turns-on early but produces a weaker signal?” What is the relationship between diode current and weak signal detection?
The next step was to express the relationship mathematically by calculating the ratio between the diode’s Vf and the level of diode current (Id) measured in the previous test (Id / Vf). Each diode could then be assigned a number (Vdx) that might possibly indicate it’s true performance potential in my own system.
The diodes with the highest Vdx values would then be A-B tested under real receive conditions to see if any (or just one!) particular winner(s) might emerge … and if Vf was as critical as initially believed.
The Vdx values proved most interesting and seemed to account for some of the anomalies noted in earlier measurements with some of the higher Vdx values coming from diodes not necessarily with a low Vf. I’m hoping that this sorting concept properly takes into account both turn-on level (Vf) and current level (Id), since a higher level in either number will compensate for a lower level in the other. Vdx values ranged from 23 to 66, with seven diodes in the higher 53-66 range.
All of the 49 diode's test parameters were put onto a spreadsheet and listed in order of their Vdx value.
The highest Vdx assignment of 66 went to my 40-year junkbox resident, a JHS Sylvania 1N3655A microwave mixer diode. It will be interesting to see if it really is the best of the lot! Although it did not produce the loudest signal (Id) compared with others, its Vf turn-on was an impressive .181V and its weak-signal detection level was good although not the lowest. A couple of the UHF diodes exhibited the interesting behaviour of picking up the UHF data stream 'clicks' from my nearby wifi booster. The 1N3655A was one of them.
Diode #2, with a Vdx of 62, is a mystery diode with a very low Vf of .197V. It was slightly louder and oddly enough, dug down slightly further than the 1N3655A, which had a slightly lower Vf. Although I don’t recall specifically, I suspect the diode may have been removed from a VCR front end many years ago.
Diode #3 with a Vdx of 61 is a modern SMS7630 Schottky microwave detector diode in an SMD package. Although it did not produce a competitive level of loudness (Id) in the diode current test, its shockingly low Vf turn-on of .147V and weak-signal detection threshold were the best of all diodes tested. Before testing, all SMD diodes were mounted on small PC boards in order to attach leads.
![]() |
SMS7630 Schottky Vf = .147V Id = 9uA Vdx = 61 |
Diode #4 (Vdx of 60) is an ISS98, another modern Schottky microwave detector. I recall seeing this diode recommended for good performance in an FM crystal radio detector. Its sensitivity level was excellent.
Diode #5 (also with a Vdx of 60) appears to be a normal germanium of unknown type. I suspect it was used as an RF mixer since it was found on a small printed circuit board with three others, connected in a diode ring configuration typically seen in balanced RF mixers. It produced high current as well as good weak signal capability.
Diode #6 (Vdx of 55) also looks like a germanium of unknown type with a body striping of gray-white-green-gray. If the last band is ignored, this could be a 1N895, a UHF germanium diode. It shows the typical internal cat-whisker type of junction often seen on the 1N34 germaniums.
Diode #7 with a Vdx of 53 is marked as a ‘95481’ on a green body. It had excellent sensitivity and produced a strong signal (Id), elevating it to the top tier to be looked at more closely.
![]() |
'95481' Vf = .246V Id = 13uA Vdx = 53 |
Diode #8, another germanium mystery, earned a Vdx of 49 due to its fairly high Id level.
The rather beat-up looking Diode #9 is marked with what appear to be house numbers, '1846' and '6628'. I believe this was pulled from an old portable radio's FM section many years ago. Interestingly, like some of the UHF mixer diodes, '1846 / 6628' detects my high speed modem data stream clicks. Additionally, this tortured specimen produced the highest level of signal among all 49 diodes, with an Id of 14uA.
Diode #10 appears to be the brother of Diode #8 with a Vdx of 48. Although it has a lower turn-on point and was a better weak signal detector, it did not produce as much Id as its sibling, dropping it one notch lower on the list. Like its brother, it also has the mystery 'T' marking. Both are most likely unmarked 1N34As.
![]() |
Vf = .252V Id = 12 Vdx = 48 |
As well, three other diodes garnered my interest. Although they ranked lower than I expected, all had previously been found to be good detectors in my system. Their lower ranking may be a hint that my system of grading is not a valid method of determining best performance. All three will be given a harder look in the upcoming elimination tests.
The first is the germanium FO-215. Often touted as 'the holy grail' crystal radio diode but I have never found it to be particularly outstanding. Maybe my system has a lower Q than it really needs in order to show its stuff. This diode is shown on the bar graph above as #11. During testing, it appeared much less capable of weak signal detection than most others but its low Vf and high Id elevated its overall ranking.
The second diode is the Soviet-era D18, a military-grade germanium in a glass '50s-style package. I have previously found it to be a very good detector but its high turn-on level lowered its ranking. The D18 appears on the bar graph as #12.
The third diode is a vintage Sylvania 1N34 from the 50s and likely one of the first 1N34s to be manufactured. Although it produces a loud signal, its Vf was higher than expected. As I recall, it was salvaged from an old parted-out Heathkit. It appears on the bar graph as #13.
![]() |
Vf = .335V Id = 13uA Vdx = 39 |
As mentioned earlier, one can measure and calculate a large amount of data for crystal diodes while they sit passively on the bench but they really need to be mounted, tested and compared in the actual system in which they will be used. Comparing diodes 'A-B' style in real time with weak signals may be better than any measurements made on a diode being bench-tested.
Will a new ‘holy-grail’ emerge from the pile? This type of testing requires a lot of careful listening so time will tell.
Testing will be ongoing over the summer / fall months ... stay tuned for the final results, hopefully in time for the fall DX season!
No comments:
Post a Comment