High harmonic content

[gkasprow]

One of the users reported that Booster has high harmonic content. Here are the measurements at 150 MHz Signal, (-18 dbm). Signal was generated by Urukul

Comparing with MiniCircuits amp

The Booster was optimized for efficiency, not for linearity. Let's see what we can do about it

[gkasprow]

So, we can make a special version of Booster, dedicated to high-linearity applications. If we go for let's say 2x QPA2237, we would be able to significantly reduce the harmonics. With 360mA of Id, we can try to beat the ZHL-2W. Note that my measurements were at 22dBm, but harmonics were in fact 20dB higher due to the 20dBm attenuator. ZHL-2W has -23dBm (after correction by 20dB due attenuator) of 2-nd harmonics at 22dBm output power.

[gkasprow]

RFPD3580 is another interesting option, we would avoid connecting two stages in parallel

[gkasprow]

I ordered a few push-pull wideband amplifiers, will give them a try.

[hartytp]

@gkasprow thanks!

So there is an obvious tradeoff: harmonics vs Idd.

If that's the situation here then I agree there isn't a design issue so much as a deliberate choice made. We wanted something with high power consumption for cases where linearity isn't particularly important. As you say, there is a reason this dissipates much less power than an equivalent MCL amplifier; we pay a price in terms of linearity.

This should be clearly communicated to users of Booster. If people want something with better linearity then we can look at designing a replacement RF circuit. First though it would be really helpful to have a clear spec for what is actually wanted from the design (bandwidth, max power, etc). I suspect that taking the current design but aiming for more linearity won't actually provide the optimal solution for people

[hartytp]

I cannot set a higher current than 250mA.

Why can't you go higher than 250mA? What limits this?

[gkasprow]

Right. We can make an assembly variant with NPA1007 which has a similar package and pinout (only bias enters separate pin). In this way the user can choose high efficiency of low distortion mode by just changing the Id.

[hartytp]

what's the downside of using NPA1007 as standard and just using a lower current by default?

[gkasprow]

250mA seems to the max due to thermal dissipation. I damaged the FET after running with a higher current. We are also limited by Booster 28V LDO - it trips.

[hartytp]

250mA seems to the max due to thermal dissipation. I damaged the FET after running with a higher current.

Okay, so 7W quiescent max.

I damaged the FET after running with a higher current. We are also limited by Booster 28V LDO - it trips.

Okay, good to know. IIRC this was a design choice by us in setting the foldback limiting resistors; the LDO itself can supply a lot more, but I may be wrong...

[gkasprow]

Okay, good to know. IIRC this was a design choice by us in setting the foldback limiting resistors; the LDO itself can supply a lot more, but I may be wrong...

exactly.

[gkasprow]

I tried the CATV amplifier. It works at 28V and provides -40.3dBc of second harmonics and -47dBc of third harmonics at 30dBm of output power. At 20dBm of output power the second harmonics is -45dBc and the third one is at -60dBc. The amp is a popular 8$ CATV hybrid. It's no longer produced but I could buy it quickly It consumes 530mA at all power levels from 0 to 30dBm. It does not need any biasing. Just 28V. I ordered another one with similar specs and the same package from Mouser but these days it takes 2 weeks until they ship it... Anyway, it looks promising. Such amp would use the cooling potential of Booster mechanics :)

[gkasprow]

It seems feasible to make such an amp compatible with existing mechanics.

[gkasprow]

some photos power stage S21 power stage S11

[gkasprow]

This particular hybrid has the potential to work at 12V. I mean simple EEM RF amp idea It consumes 270mA and at 30dBm has -30dBc of second harmonics. But the gain drops to 10dB At 20dBm it has -41dBc of second harmonics at 18dB of gain. Probably not worth it since it's out of its working conditions.

[hartytp]

The amp is a popular 8$ CATV hybrid. It's no longer produced but I could buy it quickly

How practical is it to build new products based on obsolete CATV amps? They are really nice amps, if you can get them

[gkasprow]

How practical is it to build new products based on obsolete CATV amps? They are really nice amps, if you can get them

If you keep a stock of a few hundred, that makes sense. I ordered the available equivalent (RFPD3580). Similar spec, same package, and voltage.

[gkasprow]

Another amplifier I ordered from Mouser 2 weeks ago just arrived. Now it takes even 10 days to complete the orders. It offers slightly higher power, consuming 0.5A of idle current. output 2-nd harmonics 20dBm -40dBc 30dBm -34dBc 34dBm -30dBc 35dBm -26dBc

[gkasprow]

I integrated the amplifier with the Booster RF board. I removed the attenuators and the first stage because it made the 2-nd harmonics worse. I left only PHA-1+. The overall gain is 30dB which matches Urukul well. And now this is what I get: at 20dBm, the second harmonic is at -48 dBc at 30dBm, the second harmonic is at -40.5dBc at 33dBm, the second harmonic is at -37dBc at 34dBm, the second harmonic is at -33dBc at 35dBm, the second harmonic is at -30dBc

[gkasprow]

It's time to design new boards for Booster-HL version

[gkasprow]

And this is the new PCB. The amp needs a simple aluminum adapter to dissipate the heat. It can also be screwed directly to the wall, but that would break compatibility with existing mechanics.

[hartytp]

It's time to design new boards for Booster-HL version

@gkasprow before we spin up any new hardware, can you have a look at https://github.com/sinara-hw/Booster/issues/375

[hartytp]

@gkasprow we should also look at the slow (ms) power transients (https://github.com/sinara-hw/Booster/issues/368) with the old/new amp before producing more hw

[alandoak]

Protection diodes on the input and output of an amplifier are generally a bad idea, they are non-linear elements that create harmonics and intermod distortion, even well below their voltage knee. Diodes are designed and marketed for datalines and such, not RF signals. For RF equipment that might be exposed to lightning strikes (outdoor antennas, CATV, etc.), protection usually involves shunt quarter-wave stubs, lumped element filters, and sometimes gas discharge tubes (GDT).

Also, once a component is assembled onto a circuit, the ESD tolerance is greatly increased by filtering, decoupling, trace capacitances, etc. For a lab amplifier like this: remove the diodes, slap a sticker onto the front panel that says "+24dBm max input", and maybe add a lowpass filter to the output.

[gkasprow]

@alandoak you are generally right. But in the case of this amp, we don't really care about the linearity. Without these diodes, we observed several serious damages to the first, second, and third stages. You can find all the history of the debugging in issues, which ended up with the introduction of the TVS diodes. Thanks to these diodes, the Booster is bullet-proof. Why do I think you are not right in this case? There is a Booster-HL version that has very low distortion even though all the diodes are in the same places. The only difference is the power stage. https://github.com/sinara-hw/Booster/issues/385

[gkasprow]

During debugging of nonlinearity, the TVS were the first components I removed :)

[gkasprow]

It's not an issue, it's a feature. Closing.