Voltage controlled phase shifter for 125MHz?

Hello, I'm looking for a voltage controlled phase shifter for operation at around 125MHz. RF bandwidth only needs to be a few MHz or so (more is fine), but I'd like to keep insertion loss consistent vs phase shift (maybe <3dB over its full range). Need a phase shift range of up to 90 degrees. Need something that can be board mounted (no cable modules).

I've looked around at all the companies I know, but the only thing I could find is the JSPHS-150 from minicircuits (which looks like it work fine, but it's actually much more than I'm asking for, and is pretty expensive at S30). Board mount phase shifters in this frequency range seem to be pretty rare.

I'm totally open to making my own design as well; in fact I'd prefer it that way. After doing some googling, the best resource on phase shifter design I've found is this, where it goes over a few different topologies, but very little information on their actual performance (phase range, S21 dispersion vs phase, etc). Are there more sophistocated techniques for achieving performance like the minicircuits part, such as adding more stages with smaller tuning range per stage?

Thanks in advance!

Hi, see the attachment.
Hope this can help you.

I've seen phase shifters implemented using VCOs and phase detectors in feedback, but I explicitly want to avoid using a VCO because I presume such a system has an unbounded range (since the VCO is effectively an integrator with regards to phase), as opposed to a passive phase shifter which has a bounded phase response. Also I need high control bandwidth, but not excellent linearity of phase vs control voltage. Ultimately I plan on using the phase shifter inside a larger control loop (something like polar feedback), hence why I need high bandwidth (like several MHz).

But that paper is still interesting and I have a couple questions about it. Right off the bat, I don't see how the first equation works out. Doesn't the phase locked loop force the output of the VCO and the PSD input to be both phase and frequency locked, through the action of the control loop? Is the control loop or phase detector tampered with by the control voltage in order to generate a phase offset somewhere? Also their schematics don't actually show a control voltage, but rather an output voltage, as if it's a phase measurement system, not a phase shifter?

How about using an inductor with a varicap diode in series. The phase at the point in the centre will change. If you have an op-amp, with negative feedback, you can keep the amplitude constant, and have that set by a couple of resistors. I've never tired this I would add - it just seems like it might be an approach. I have used a commercial phase shiftier before, but that was loads of money - far more than the S30 you find too expensive. But I only needed one, whereas I expect you are looking at something for production, where cost is important.