Noise in "Jelly Bean" Regulators - Part 1

In the course of investigating the retrofit of a low noise RF pre-amp to an old 2m mobile, RT85, a random thought from nowhere made me stop and consider regulator noise. 

My goto part for 3 volt regulators in microcontroller projects is the LN6206. Cheap and reasonably low quiescent current. However, there is no specification of the output noise voltage. 

Since I had seen a TL431 on a PCB earlier in the day the device was foremost in my mind. I wouldn't need a pre-regulator to reduce the voltage to the LN6206 which is a 6V maximum input device. However, the TL431 datasheet I had also made no mention of noise.

After a bit of searching I became cynical of the claims the TL431 is low noise. I found just one objective assessment of noise which helped reassure me. Going from 50 nV/√Hz to noise in a 0-100kHz bandwidth and adjusting for gain from the 1.25V band gap voltage of the TL431 to 3V returned 38uV. Which is almost identical to the 40uV output noise voltage of the L78L33C. 

I'm not sure that qualifies as low noise, since a LT3094 is specified as 2 nV/√Hz but costs over 50 times more! I note in the web page above that an additional capacitor could be expected to reduce the output noise voltage considerably. Back of the envelope suggest under 10uV.

Can we do better? It appears so. A simple one transistor circuit after the regulator should make a large difference. 

Verification is the hardest part. I could not find a low noise amplifier circuit that I could re-create with my junk box. My attempts to build a low noise amplifier in a shielded box are yet to bear fruit. 

My whole approach to electronics is to re-use previous work until the need for improvement arises. While trawling through the internet I saw numerous statements that low noise regulators are vital to oscillator performance. This is not something I had considered beyond decoupling supply lines. So it appears the need for improvement exists!

73's

1284MHz PLL for 6cm Receive Converter

 I was recently fortunate enough to be given a Mitec Down Converter by VK6UM. 6cm but with a 140MHz IF. 

No problem, or so I thought, let's replace the 1280MHz oscillator with a PLL design generating 1284MHz. That would give a 144MHz IF.

I took a 23cm beacon  board previously described and hacked it around. 

First issue I faced was the PLL lock detection. I used the same MB1507 PLL chip but I increased the reference frequency to 500kHz. I had a few Forth commands that I could use over the serial link to change the frequency. While the VCO tuning voltage would shift according to the frequency I set, the lock detector never went into a lock waveform. After a few days I tried a 250kHz reference and straight away we had lock. 

Some qualitative testing of the PLL output was needed. After a bit of head scratching it occurred to me I could take a 800MHz output from my GSM test set, mix it with the 1284MHz signal, and look at the 484MHz result on the GSM's spectrum analyser. That worked and revealed some 250kHz spurs. I was able to reduce those with another rework of the loop filter values.

The next issue I faced was microphonics. Replacing the loop filter capacitors with tantalums as necessary reduced the level of microphonics when I tapped away with a screwdriver. Listening with an ICQ7a on narrow band FM to the 1284MHz signal was like a road crew sound check. Tap tap TAP TAP.... Useful test but I later realised I had forgotten to replace a 10uF MLCC on the regulator output feeding the VCO. By then it was all boxed up and installed. 

So the microphonics are there for now. I modified the PLL board based on what I had learned and in due course will rebuild the 1284MHz PLL. Hopefully when Larry, VK6UM, checks out the result it will be serviceable.

73's

Relay Life for Relay #2

After some 10 days of continuous operation the test board flash memory was full with over 2 million cycles of data at a constant 2 Amps using a constant current generator. Below is a picture of the actual relay, nearly destroyed when it was removed from its original PCB.

 

I chose this recovered relay because it was damaged. It was surprising that despite the abuse during removal the results, below, show there was no real degradation over 2 million cycles at the rated contact current capacity. Perhaps 30 volts at 2 Amps would give a different result compared with 12 volts at 2 Amps. But for now I am completely satisfied that this relay, of which I have recovered dozens, will be a suitable basis for use as a 50 watt 6m changeover relay.

 

To put this in context the ADC returns a reading of 214 for 0.685volts. For each cycle 16 readings (the graph was wrongly label as the sum of 8 readings) of the voltage across the relay contacts is taken, summed, and stored for the analysis. So a sum of say 200 means there was 40 mV across the terminals. With 2 Amps passing through the contacts that means the resistance was 20 milli-ohm (0.020 ohm).

That is probably enough from me on the subject of relay life. I'm satisfied that operated within the specified ratings a relay, even when recovered, is not going to wear out. All you have to watch out for is  relays re-printed with over-stated specifications sold by the Chinese on Ali-express etc. Similar to the ludicrous ratings printed on 18650 lithium batteries they sell.

73's