Tuesday, 6 October 2020

IF Crystal Filters - NDK Type numbering decoded?

I wanted to know more about the crystal fitlers used in the Tait Orca handhelds ie 45E12AA and 45E15AD. 

I could not find the specifcations for these exact filters. However, after looking at some NDK datasheets I believe these crsytal filter are as follows:

 45 - denotes the nominal center frequency

E - couldn't work the meaning out. Perhazps E means 2 poles and EE means 4 poles?

12/15 - bandwidth inkHz

A  - finalises model number

A/D at the end denotes the center frequency.


So, 45E12AA apears to be a 45.1MHz 2 pole crystal filter (seaching for 45E12A did not return a model datasheet, perhaps a special Tait specification?)

and 45E15AD is a 45.1MHz 2 pole crystal filter (search for 45E15A to see the model datasheet)


Seems strange that the center frequency is not denoted by a consistent letter.  But as expected when looking for relatively old datasheets, not everything is on the web. 

Best to keep the crystal used to mix the IF down to 455kHz with the filters so you can derive the center frequency if no data can be found.




Monday, 28 September 2020

23cm FM Receiver - Musings

 All the assemblies for a 23cm FM receiver are coming together i.e.

Once the 1296MHz signal is mixed down to a nominal 446MHz I still have not decided on my approach.

 I am leaning towards the last option, the RDA5807.

I have a number of Tait UHF Orca handheld boards which I could either use as donors for the necessary filters and IC's, or as a working IF strip. These boards probably have faults but after cutting to size I have enough boards to be able to get one working. With a bit of work I could also use the synthesizer to generate the approximate 400MHz signal needed to mix down to the 45MHz IF strip. But this all becomes a bigger project than I had contemplated and the risk is not so much technical as motivational. Will I ever get it finished?

The simplest, most expedient, lowest risk is to secure a working handheld inside the case and move forward. But that does not sit comfortably with me.

The left field final option is the RDA5807. While walking back and forth to the workshop it dawned on me that this single chip FM Receiver would be a real time saver if it worked. I have some experience with these and the risk relates to how well it will work with a narrow band FM signal. If I can get sufficient recovered audio with acceptable noise levels this will be the quickest, most expedient way forward. 

Time for some testing.



Friday, 11 September 2020

Brake Cleaner used as a PCB Flux Cleaner

The usual safety disclaimers apply. Don't drink the stuff like some loopy folk do with hand sanitizer
( good one Mr President), don't inhale (didn't a past President make some claim concerning pot?) and watch the naked flames. It's your shack, your health and your life. Take it seriously because it is totally your responsibility.

In my ham projects it's ok to take shortcuts. You should never try this on a commercial product you sell. You have been warned!

I haven't been satisfied with methylated spirits as a flux cleaner for some time. And isopropyl alcohol is expensive and not that effective in my experience. I was ready buy proper flux cleaner like that I use in my own business but somewhere in my internet browsing I came across reference to using brake cleaner as a flux cleaner. Never shy of testing such claims I thought I'd give it a whirl.

Dropping into the helpful people at Repco, and being up-sold by and extra $1.20 to their premium brake cleaner costing $8.29, I came home and trialed the brake cleaner.


Even on my personal projects I'd only use this in a pinch. But it works out to be substantially cheaper than proper flux cleaner and is readily available. Not too many of us have an electronics outlet on our doorstep open on a weekend.  But the long term consequences on the board are unknown and it leaves more of a film on the board surface than proper flux cleaner. Other brands might be different though.

I devised a test for three flux residues :

  1. Rosin from a homemade flux (rosin dissolved in methylated spirits and wiped on with a cotton wool tip or splashed on the solder reel and allowed to dry on the solder wire - why you would bother with the crap some Chinese sellers pass off as solder is beyond me, but it might make it usable)
  2. residue from Multicore solder
  3. the flux unused from a cheap syringe type product, again from China (works for me though)

Fluxes 1 and 3 are the result of trying to make that cheap solder flow nicely over the joint. Personally, I keep it on hand to give away to people to educate them on why saving cents can ruin your day. But enough China bashing for one day!

A quick test board was prepared using a rejected board I had to hand and is shown below:


The rosin flux and Mulitcore flux have proved difficult to remove in the past. The cheap eBay flux can be effective but is impossible to apply in small amounts so it is always messy.

Three brief spurts with the brake cleaner and I had this:

Rosin residue completely gone!

Multicore residue almost gone.

Cheap syringe flux partially gone.

I was starting to get excited now. A tiresome chore solved so easily. A quick dab of brake cleaner on a cotton wool bud rubbed on the remaining residues and the resulting board:



The soldermask did not appear to suffer and the result was a spotless board. 


In conclusion it appears brake cleaner does the job very well in the short term. The long term consequences are unknown and that needs to be considered. How comfortable are you that the construction effort today might not last?




Friday, 4 September 2020

23CM 500mW Amplifier AH101 - duplicated with revised matching

Since I have a lot of these boards I used one to make a test fixture. OSL calibration of the VNA right at the device legs proved feasible since I have some 50ohm RF chip resistors. Fortunately, the measured S parameters were broadly consistent with the tabulated S parameters in the datasheet.

With the device soldered onto hte board I switched between the calibration at the device legs and the calibration at the SMA sockets. This allowed me to infer the electrical length and impedance of the connecting micro-strips, and any strays.The length and impedance were close to the expected value so I was able to confidently rework the matching. 

As you can see below I was able to use a single inductor at the SMA end of the board for the input. I inferred that a microstrip with slightly higher impedance would give a better match. The crazy cutter went mad and the result was just a few dB improvement to a return loss of 21dB. I overshot the mark!

Matching for the output network was placed at the device end of the output microstrip. The capacitor to ground was meant ot be 1.5pF, but I found the slightly smaller 1pF capacitor gave me a better result. Return loss of 24dB was achieved.

That makes two successful amplifier boards and a great learning exercise. I am a lot more confident with the Smith Chart now and RFSim99.  I still have a heap of the AH101 devices so one day I would like to try 4 in parallel using Wilkinson dividers.



That might have been the end of the matter but I noticed the claimed specifications in the datasheet were not supported by the S parameters. Which should be embarrassing to a manufacturer!

To be fair, this is a discontinued part and the original manufacturer has changed hands several time. So let's look at the TQP7M9102. This is Qorvo's suggested replacement. What we find is a table of S parameters in the datasheet explicitly qualified by: 

"Test Conditions: V CC  = +5 V, I CQ  = 135 mA, Temp. = +25 °C, unmatched 50 Ohm system, reference plane at device leads"

The S parameter file downloaded states:

"The data is for the unmatched device in a 50 Ohm system, 2-port file de-embedded to the device leads."

So my expectation is this should be the same dataset. The best I can say is they are broadly similar. I drew this to Qorvo's attention. The response simply acknowledged the error. With that attitude I can no longer consider Qorvo for commercial applications which means it is unlikely any of their products will ever find a way into my ham projects. Once those AH101's are gone that's it!



Friday, 21 August 2020

23CM 500mW Amplifier AH101

 I have not determined why the SXB4089Z input matching was out. After lots of maths and component substitution I suspect the mmic itself is damaged. I will swap it out soon to confirm this.

In the meantime I tested an AH101 device. This was made awkward because the test board was set out with matching networks for the AH1, which are quite different. With some hacking I got it to work. Input return loss was initially 17db but I was able to tune that to under 30dB with a little foil gimmick capacitor. Output return loss was 17db. Gain was 13dB.



Friday, 14 August 2020

23cm 500mW Amplifier SXB4089Z

 After calming down over JLPCB's appalling process for dealing with their errors, which is to deny everything and obfuscate to the point I gave up, I populated one of the test boards for the SXB4089Z. My earlier posts here, here and here set out the steps I took to end up with the following:


My results did not quite match this. Using the nanoVNA to measure the input and output ports, adjust for the attenuation used and then combining everything into a single file for RFSIM99 to read showed that the output matching was nearly spot on (I will need to change the 1.2pF output capacitor to 1.0pF) but the input matching was all over the place. I repeated the measurements and got the same result.

I am confident that once the input reflections are resolved that it will only take a small tweak to get this working well.



Friday, 7 August 2020

New part - SMA3109 MMIC

Something different....

I stumble across new parts from time to time that set a price / performance benchmark. So I thought it would be worth alerting readers to these parts. Here is the inaugural post on this subject.

The new benchmark in my shack for general purpose mmics is the SMA3109. It happens to be the cheapest mmic that LCSC stock. I had wanted to use this commercially but the initial lack of S parameters meant I passed it over in favor of another device. That decision might get revisited.

Despite the low price the SMA3109 offers outstanding performance from VHF through to 4GHz. Compare it with other low cost mmics such as the BGM1013 and you will see why I rate it so highly.

When I evaluate a MMIC I consider:
  1. stability
  2. input and output return loss
  3. noise figure at intended frequency
  4. gain
  5. ease of hand soldering
This device is ticking all my boxes but the NF of 4dB rules out demanding low noise applications.

I suspect it is useful outside these ranges but the S parameters kindly supplied by Onsemi technical only cover 100MHz to 4GHz.

 Band     Gain S11     S22
 2m     23dB     -28dB     -13dB    
 70cm     24dB -24dB     -16dB
 23cm     24dB -38dB -16dB
 13cm     23dB -15dB -14dB
 9cm 21dB -11dB -17dB

The input match on 23cm is outstanding and I will definitely use this device in my 23cm projects where filters are connected to the input. For driving filters I would consider at least a 3-6dB pad to provide a better source termination. Being very stable across the entire frequency range matching if needed is straightforward.

Where linearity is important I would budget to get around 0dBm out of the device.

My hunch is that the feedback within the device makes it useful below 100MHz except for the output becoming a poor match to 50 ohms.

The device appears to be very useful from 2m through to 9cm and represents terrific value for signal outputs up to the Po(1dB) level of +4dBm. Definitely worth a look at US 11cents each!