|
QUOTE (from the first link):?
They don't blow up in an extroadinary fashion, but instead will draw
maybe 4x the current (from about 5mA ->20mA) in their failure
condition. They also exhibit low gain whenever they reach this
condition.
High current draw is a certain indication of parasitic oscillation!??
A useful tool in damping any parasitic oscillation (no, this is NOT "snake oil"!):? lay a finger firmly on the '590.? The current will drop immediately if there is an oscillation.? Your finger is introducing enough loss to quench any oscillation.? Be sure to do this before the device signs its own death certificate - rapidly after observing the overcurrent condition.? If current drops with the "finger test", immediately turn off power and do what you can to alleviate the oscillation (small series resistor on the collector).?
In dealing with these high ft devices, I would strongly recommend a set of pads be added to the PCB to accommodate an SM resistor between 22 and 68 ohms placed in series with the collector.? If other devices are used with lower ft's, the pads can always be loaded with a zero-ohm jumper.
In addition, these devices are extremely sensitive to ESD damage.? Be careful in handling, placing, and soldering.? Generally once they are in circuit, things are much more robust.??
Dave - W?LEV?
On Fri, Jan 31, 2025 at 8:23?PM Dhiru Kholia via <dhiru.kholia= [email protected]> wrote: Thanks for all the tips, info and help!
I have access to TinySA Ultra but I believe monitoring my power supply
for unexpected current consumption might work better to detect
oscillations.
I was reading the following page earlier today:
...
I am attaching the PCB files to this email to aid the review process.
The complete set of files is open at the following link:
Cheers,
Dhiru
|
I've just taken an in-depth engineering purusual of the NXP datasheet for the BFU590Q:
1)? For a BJT with an ft of 8.0 GHz, the gain sure drops off rapidly with frequency!? They don't even rate gain above 1.8 GHz.
2)? And no data is given for the noise figure.
3)? It is rated for ± 2kV for ESD.? But, believe me, that's not much.? That can easily be accumulated in the winter on a dry day or even summer on a dry day.?
The device is certainly aimed at the 900 MHz and 1.8 GHz cell bands.
Dave - W?LEV
QUOTE (from the first link):?
They don't blow up in an extroadinary fashion, but instead will draw
maybe 4x the current (from about 5mA ->20mA) in their failure
condition. They also exhibit low gain whenever they reach this
condition.
High current draw is a certain indication of parasitic oscillation!??
A useful tool in damping any parasitic oscillation (no, this is NOT "snake oil"!):? lay a finger firmly on the '590.? The current will drop immediately if there is an oscillation.? Your finger is introducing enough loss to quench any oscillation.? Be sure to do this before the device signs its own death certificate - rapidly after observing the overcurrent condition.? If current drops with the "finger test", immediately turn off power and do what you can to alleviate the oscillation (small series resistor on the collector).?
In dealing with these high ft devices, I would strongly recommend a set of pads be added to the PCB to accommodate an SM resistor between 22 and 68 ohms placed in series with the collector.? If other devices are used with lower ft's, the pads can always be loaded with a zero-ohm jumper.
In addition, these devices are extremely sensitive to ESD damage.? Be careful in handling, placing, and soldering.? Generally once they are in circuit, things are much more robust.??
Dave - W?LEV?
On Fri, Jan 31, 2025 at 8:23?PM Dhiru Kholia via <dhiru.kholia= [email protected]> wrote: Thanks for all the tips, info and help!
I have access to TinySA Ultra but I believe monitoring my power supply
for unexpected current consumption might work better to detect
oscillations.
I was reading the following page earlier today:
...
I am attaching the PCB files to this email to aid the review process.
The complete set of files is open at the following link:
Cheers,
Dhiru
--
|
I use a Siglent SSA 3021X SA for checking for oscillation. Reading that eevblog thread certainly would lead you to believe the BFU590 is hard to tame and readily prone to oscillation, yet as I've shown in my ugly build there is no oscillation at all for this particular configuration, at least using the QX version of the transistor. And putting a -61 material small ferrite bead in the collector lead will generally stop oscillation, in my experience.
Steve AA7U
toggle quoted message
Show quoted text
On 1/31/2025 10:46 AM, Dhiru Kholia via groups.io wrote: Thanks for all the tips, info and help!
I have access to TinySA Ultra but I believe monitoring my power supply
for unexpected current consumption might work better to detect
oscillations.
I was reading the following page earlier today:
...
I am attaching the PCB files to this email to aid the review process.
The complete set of files is open at the following link:
Cheers,
Dhiru
|
On Fri, Jan 31, 2025 at 08:40 AM, Fred M wrote:
I believe the CMRR measurement in a 50 Ohm environment is not the same as the CMRR from a Hi-Z reactive source impedance
I agree the two environments will lead to different results.
?
From my reading, there are those are concerned that the long run of coax is picking up indoor interferences which subsequently appears as a voltage between earth and the dual input terminals. Perhaps, the 50 Ohm environment (per the official standard) is more suitable for their CM environment?
?
There is another official setup which I think is useful for testing CMRR and is reflective of the above condition. EN55020 (BS905 to Martin :-) ) coupling unit type A which injects the test signal into the coax shield. /g/loopantennas/photo/300229/3881311
I'll offer a tried-and-true "most-of-the-time" solution: ?Instal a 1/4 or 1/8-watt - better yet, an SMD - 22 to 68 ohm resistor in series with the collector/drain. ....
In addition, you should have a spectrum analyzer or other means of verifying there is no oscillation at or below the rated ft.
?
Hi Dave,
Not all instability can be solved by resistive series stabilization. Some require resistive shunt. A stability analysis using the Smith chart is necessary to find out which one can work. The below tutorial is helpful.?
http://www.n5dux.com/ham/files/pdf/Avoiding%20RF%20Oscillation.pdf
?
When I only had a 1.8 GHz HP8591 spec analyzer (at work), it was the thermocouple power sensor that clued me in to out-of-band instability. The sensor can respond to frequencies way above its specified maximum, though un-calibrated.
?
To be honest, I'd thought that adding a 1:1 transformer on the amplifier input would have made a fairly large change, by improving both the isolation and balance, but no matter how much I messed around with the configuration, I couldn't make it any better than the same amplifier, without a transformer input.
?
Hi Martin, how about experimenting with modifying the LZ1AQ into a differential amp with a shared emitter resistor? I think the output common emitter stage are readily amenable to this modification. However, the input common base, hmmm.....
?
I have access to TinySA Ultra but I believe monitoring my power supply
for unexpected current consumption might work better to detect
oscillations.
?
Yes, it's true that, in general, the current will increase when instability occurs. However, one must have prior knowledge of the current of a stable unit in order to know if the measured current indicates instability. Moreover, I have encountered the opposite, where current decreases during instability. The oscillation cause negative polarity rectification in a connected diode (could be a ESD protection diode in the transistor or active bias circuit) which slightly decreases the current consumption. So, I find current monitoring unreliable - at least for me.
?
?
?
|
I think Fred's suggestion of an E-Field monopole simulation is still valid.
?
Although the coax may be longer providing a better ground path, the size of the loop providing the antenna element, and its value of equivalent capacitance remains the same.
?
Conducted interference, and ground loops via the feed coax are different factors, which require separate analysis.
?
Regards,
?
Martin
?
?
On Sat, Feb 1, 2025 at 03:07 AM, <biastee@...> wrote:
From my reading, there are those are concerned that the long run of coax is picking up indoor interferences which subsequently appears as a voltage between earth and the dual input terminals. Perhaps, the 50 Ohm environment (per the official standard) is more suitable for their CM environment?
|
I've forgotten about the small ferrite bead.? You can also pass both the emitter and collector (insulated from eachother) through the same bead.? This acts as a common mode choke and forces equal currents to be drawn by both elements of the BJT.? THis is also a good oscillation quencher.
Dave - W?LEV ?
On Fri, Jan 31, 2025 at 9:22?PM Steve Ratzlaff via <ratzlaffsteve= [email protected]> wrote: I use a Siglent SSA 3021X SA for checking for oscillation. Reading that
eevblog thread certainly would lead you to believe the BFU590 is hard to
tame and readily prone to oscillation, yet as I've shown in my ugly
build there is no oscillation at all for this particular configuration,
at least using the QX version of the transistor. And putting a -61
material small ferrite bead in the collector lead will generally stop
oscillation, in my experience.
Steve AA7U
On 1/31/2025 10:46 AM, Dhiru Kholia via wrote:
> Thanks for all the tips, info and help!
>
> I have access to TinySA Ultra but I believe monitoring my power supply
> for unexpected current consumption might work better to detect
> oscillations.
>
> I was reading the following page earlier today:
>
>
>
> ...
>
> I am attaching the PCB files to this email to aid the review process.
>
> The complete set of files is open at the following link:
>
>
>
> Cheers,
> Dhiru
>
>
>
>
>
|
For what it's worth, we have found that Vcc to an antenna-mounted preamp (or in our case, the varactor control tuning bias) must be ultra clean. SMPS have no place there. A a GOOD SMPS will have upwards of 100+ uV of switchcing noise and at least 30uV of hash ... very typical. As suggested, at this point in a receiving system, where we deal with uV incoming signals, sub-millivolt noise on the Vcc is mandatory, otherwise all the concern about transistor type selection, shielding, etc, is for nothing. As suggested, a decent linear supply with a regulator reduction of at least 4-6 volts (18 to 12, for example), with bypass caps on both, in and out, legs, like .001uF and followed by a cap-input pi-filter resonated to something down to 2-digit Hz will give you almost battery-pure DC to feed the preamp (something like 10uH and 10uF) It's not only the conductive line level noise that ruins the S/N factor, but also the radiated junk RF that is transmitted from the SMPS to the antenna. With strong signals the difference probably won't be noticed. We built some proof-of-play systems for AM b'casters who deal with double digit uV and would never know the difference, but if you need to recover those 1 uV, or less, signals then add the few extra components for Vcc noise levels that usually run about 1/3uV or less. It will help you sleep at night knowing you put your best effort into your LP.
Bob, N1KPR AmComm
-
- Youtube: N1KPR
- Youtube: Ham Radio Doctor
Engineering, where enigma meets paradox
On Saturday, February 1, 2025 at 01:56:01 PM EST, W0LEV via groups.io <davearea51a@...> wrote:
I've forgotten about the small ferrite bead.? You can also pass both the emitter and collector (insulated from eachother) through the same bead.? This acts as a common mode choke and forces equal currents to be drawn by both elements of the BJT.? THis is also a good oscillation quencher.
Dave - W?LEV ?
On Fri, Jan 31, 2025 at 9:22?PM Steve Ratzlaff via <ratzlaffsteve= [email protected]> wrote: I use a Siglent SSA 3021X SA for checking for oscillation. Reading that
eevblog thread certainly would lead you to believe the BFU590 is hard to
tame and readily prone to oscillation, yet as I've shown in my ugly
build there is no oscillation at all for this particular configuration,
at least using the QX version of the transistor. And putting a -61
material small ferrite bead in the collector lead will generally stop
oscillation, in my experience.
Steve AA7U
On 1/31/2025 10:46 AM, Dhiru Kholia via wrote:
> Thanks for all the tips, info and help!
>
> I have access to TinySA Ultra but I believe monitoring my power supply
> for unexpected current consumption might work better to detect
> oscillations.
>
> I was reading the following page earlier today:
>
>
>
> ...
>
> I am attaching the PCB files to this email to aid the review process.
>
> The complete set of files is open at the following link:
>
>
>
> Cheers,
> Dhiru
>
>
>
>
>
|
On Fri, Jan 31, 2025 at 07:07 PM, <biastee@...> wrote:
Moreover, I have encountered the opposite, where current decreases during instability. The oscillation cause negative polarity rectification in a connected diode (could be a ESD protection diode in the transistor or active bias circuit) which slightly decreases the current consumption.
The curious case when oscillation lowered the operating current :-)
?
The red trace is the stable condition when Idd = 75mA, while the black trace is the unstable condition when Idd = 50mA. The oscillation is ~300 MHz, and the 2nd & 3rd harmonics are shown on the graph.
?
The cause of this counterintuitive phenomenon is the strong oscillation was rectified by the active bias (which is identical to a negative polarity detector) and this generates a negative voltage which reduces the gate bias.
/g/loopantennas/photo/300438/3881635
?
|
On Sat, Feb 1, 2025 at 03:19 AM, Martin - Southwest UK wrote:
I think Fred's suggestion of an E-Field monopole simulation is still valid.
?
Although the coax may be longer providing a better ground path, the size of the loop providing the antenna element, and its value of equivalent capacitance remains the same.
?
Conducted interference, and ground loops via the feed coax are different factors, which require separate analysis.
Two different paths for noise ingress, but one common parameter (CMRR) standing in their way?
|
Although CMRR is an issue, it's not always the overriding factor.
?
Noise is frequently introduced via a poor quality power supply, badly designed bias tee, ground loops, and poor screening of the feed line, and the amplifier CMRR performance is not going to significantly change the end result.
?
I'd suggest that it's always best to check for some of the more obvious issues first, by looking at the spectrum with the amplifier powered, but with the input shorted, and no loop connected, and then fixing those problems, before you move on to try and resolve any additional contribution due to poor CMRR.
?
Regards,
?
Martin
?
On Sun, Feb 2, 2025 at 05:49 AM, <biastee@...> wrote:
Two different paths for noise ingress, but one common parameter (CMRR) standing in their way?
|
This applies to FETs, either JFETs or MOSFETs.? Especially MOSFETs have ESD protective diodes.? BJTs generally do not.? Therefore, the noted decrease in current with oscillation likely applies to only FETs.? The '590 is a BJT.?
Dave - W?LEV
Moreover, I have encountered the opposite, where current decreases during instability. The oscillation cause negative polarity rectification in a connected diode (could be a ESD protection diode in the transistor or active bias circuit) which slightly decreases the current consumption.
The curious case when oscillation lowered the operating current :-)
?
The red trace is the stable condition when Idd = 75mA, while the black trace is the unstable condition when Idd = 50mA. The oscillation is ~300 MHz, and the 2nd & 3rd harmonics are shown on the graph.
?
The cause of this counterintuitive phenomenon is the strong oscillation was rectified by the active bias (which is identical to a negative polarity detector) and this generates a negative voltage which reduces the gate bias.
?
|
I was talking about transistors in general, not specifically the BFU590. Although BJTs don't have protection diodes, they are often use active bias / current mirrors. The current mirror is another BJT with its collector and base tied together and the emitter connected to ground - which is effectively, a diode.
Hence, current decrease during oscillation is possible with both BJTs & FETs.
|
Thanks for sharing!
This will be my next project.
Greetings from Norway,
Egil - LA2PJ
|
My evaluation of the PA0FRI preamp.
?
:-) Critiques welcome!
73, Leong, 9M2LCL, ex-9W2LC
|
Well done Leong.
?
For more convenience. Are the images available in a higher resolution, e.g. by klicking on it?
?
regards, Fred
?
|
Hi Leong
?
With regard to: Fig. 16: Transistor models from different manufacturers exhibit huge differences in gain simulation!
?
At present there are mainly PZT2222A from Nexperia/NXP and from Onsemi in active production. Which model is closest to reality?
?
I am surprised of the gain peaks with the different SPICE-models and i do not really understand what causes them. It looks like frequency dependant resonances with external components. Any idea?
?
regards
Fred
?
?
|
On Fri, Feb 14, 2025 at 1:21?PM biastee via groups.io <biastee@...> wrote:
My evaluation of the PA0FRI preamp.
:-) Critiques welcome!
Awesome work Leong and thanks for sharing! Cheers, Dhiru
|
Hi Fred & Dhiru,
?
Thank you for your kind praises. I am standing on the shoulders of giants, namely PA0FRI & Martin G8JNJ whose loop aerial impedance measurement allowed me to create the dummy aerial.
?
For more convenience. Are the images available in a higher resolution, e.g. by klicking on it?
?
Google site doesn't have that feature. One workaround is to use the browser's Print > Save to PDF function because it has the effect of magnifying the images while keeping the text at the same size. Alternatively, let me know which figure/s that you would like to view, and I will post it/them to this group's Photos section.
At present there are mainly PZT2222A from Nexperia/NXP and from Onsemi in active production. Which model is closest to reality?
?
I am amazed by your encyclopedic knowledge of transistors! :-) What I am going to write next is heresy - the transistor model that best matches the gain measurement is ..... BC547B! No gain peak artifact, at all!
/g/loopantennas/photo/300703/3886460
However, its S22 has poor resemblance to reality, sigh...
?
Anyway, an accurate answer can only be obtained if I were to measure the 2N2222A's s-parameters in a 50 ohm fixture and then compare them to the different models. This is the only way to eliminate the confounding factors due to external components.
I am surprised of the gain peaks with the different SPICE-models and i do not really understand what causes them. It looks like frequency dependant resonances with external components. Any idea?
I am clueless with regard to the gain peaks. I have tweaked the surrounding components to see if they can affect the peak, but nothing works. So far, only damping the ideal (lossless) balun with R7 to account for the former's loss seems to work.
?
However, the absence of the gain peak in simulations with Siemens PZT2222A and BC547B seems to indicate that the gain peak is inherent in the Motorola and Natsemi models.?
?
Good discussion. I am glad for the opportunity to get the above information off my chest. :-) Previously, I have thought that nobody here gave a damn to circuit modelling.....
?
73, Leong, 9M2LCL, ex-9W2LC
?
|
Bob, N1KPR
LA2PJ - Egil