Looks very suitable. I'd advise building the transformers as shown and measure them over 200 KHz - 2 MHz. You should also research the core material used. A different choice of core material may be appropriate for your project. It might be that just changing the core material is all that is required. The original design will have chosen a higher frequency core material. Someone plotted the performance curves for a number of core materials. I'll see if I can remember where. It was some ARRL publication is the only thing I *think* I remember.
Disclaimer: I understand the principles involved. However, I do not have detailed knowledge of all the calculations, etc. I have a 5000+ volume library for dealing with that. So I swim to some new spot and in the process forget something I once knew.
I'm just now at the point where I can start serious work on my major project, FOSS FW for cheap DSOs. Though the last few days have been spent on packaging an EBay BG7YBL diode noise source. And I'm on the hook to do hookup some wash basins for my sister today.
I guess there's no excuse for not having a function generator and oscilloscope now.
Well, I did something productive: I wrote Python code that simulates the paths of ions in a quadrupole mass spectrometer.
I still need to write something to actually put this to use.
# -*- coding: utf-8 -*-
"""
Created on Wed Mar 21 23:38:33 2018
@author: ikrase
Code to simulate the paths and stability of ions in quadrupole mass
spectrometers.
This is heavily based on the reference of
http://www.chem.uky.edu/research/Lynn/pdfs/JChemEd_quads.pdf
"""
import numpy as np
import scipy as sp
import scipy.integrate
from math import *
from numpy import array
def fMathieu_qp(t, u, au, qu):
''' RHS differential equation of Matthieu for quadrupoles
u = X/Y position, combined (ODE vector)
t = quasi-time -- normalized to angular frequency.
au, qu = dimensionless parameters, functions of all quadrupole properties.
(See referenced paper)
'''
return np.array([u[1], -(au + 2*qu * cos(2*t))*u[0]])
#return np.array([u[2], - (w0 * w0 + eps * math.cos(w1*t))*u[1]])
def alt_fMathieu_qp(t,u,au,qu,freq):
'''Alternative non-combined equation of Matthieu, in strict time domain.
Freq is the actual frequency in hertz.
u is [x, x', y, y']. '''
w = 2 * pi * freq
xpp = - ((w/2)**2) * (au + 2*qu*cos(w*t))*u[0]
# at some point I'll figure out which is the squshed axis and which is stretched.
ypp = ((w/2)**2) * (au + 2*qu*cos(w*t))*u[2]
return array([u[1], xpp, u[3], ypp])
def quadrupol_test(r, m, Vrf, Udc, freq, duration=False, tstep = 0.1E-7):
'''Test whether a given quadrupole system is stable or unstable.
r: quadrupole center-to-electrode radius / inscribed radius in mm.
m: ion M/Z ratio in amu to electrons
Vrf: RF zero-to-peak voltage in volts
Udc: DC spread voltage
freq: Actual frequency (not angular) of the quadruopole in hertz
'''
rspan = r * 0.125 # Radius of input particle beam
aqbot = ((2 * pi * freq * r)**2 * 1.036427E-8 * m ) # bottom of A and Q
a = 4 * Udc / aqbot
q = 2 * Vrf / aqbot
tb = min(0.002, 2, sqrt(a),abs(2-sqrt(a)))
if tb < 1E-3: tb = 5E-2 #divide by zero and over-value guard.
tlong = 2 * pi / tb # Estimate of "natural" frequency
# (if there is one.)
if duration == False:
dur = 10 * tlong
else:
dur = duration
#oderes = sp.integrate.solve_ivp(lambda tt, yy: fMathieu_qp(tt, yy, a,q),
# (0,dur), np.array([rspan, 0]), max_step = 0.05)
oderes = sp.integrate.solve_ivp(lambda tt, yy: alt_fMathieu_qp(tt, yy, a,q, freq),
(0,dur), np.array([rspan, 0, rspan, 0]),max_step = tstep)
if not oderes.success:
raise RuntimeError("Failed to integrate differential equation")
return oderes
I'm having a bit of a hard time telling what the graphs refer to.
I think it's reflection and transmission (s11 and s21) for a 1:1 transformer (common ground, 0 degree phase), on the given core and number of turns.
Which means s11 gets you a good measure of the core's intrinsic impedance characteristic (that is, not its A_L, but more general than that, its impedance per squared turn, plotted over frequency), up until high frequencies where the transmission line equivalent of the winding dominates. Which is why the one-turn transformer has a high cutoff (off scale), whereas the four-turn transformer is just beginning to cut off at the top end.
Cutoff manifests as an equivalent series inductance (LF approximation, the plots don't get very close to the actual resonance (which is a notch, followed by another pass band) so this is an acceptable way to model it), hence the coincident drop in s21 and rise in Im(s11) or VSWR(s11).
It's a big disingenuous to call it "core loss" when the headline plot is insertion loss. One must understand what's being measured. Indeed, |A_L| is increasing with frequency (for a core this size, typically until maybe 100MHz), so we should expect transformer insertion loss to be decreasing (i.e., a higher parallel impedance, loading down the signal less and less). You take s11, convert it to Z_core (separating Z_winding if necessary), divide by frequency to get [complex] inductance, then divide by N^2 to finally get [complex] A_L.
Tim
I was at the warehouse today. I didn't find any high power coil sets, but I did find a big pile of these. They are the RF coils from an RU-12 aircraft receiver. This particular set is marked 850-1330 KHz. They are plugins, so you might construct the RF amplifier with a set of two or three to cover the desired range. I didn't measure the wire, but I'd guess 26-28 gauge. So you might need to wind a new coil on the form with heavier wire. The coil forms appear to be phenoilc.
Yours for postage if you're interested.
At least for now I am saying "Thank you, but no thank you".
None of the stuff is going anywhere any time soon. I do think you should consider the pluggable coil set concept. That would greatly simplify spanning 600-2000 KHz with high Q.
Offer stands for anyone else with a good use for some of these.