Beginner attenuators set 3dB, 6dB, 10dB, 20dB

[VSV_electron]

Hello,

I'm getting myself a TinySA Ultra compact spectrum analyzer for my radio experiments and I've seen a few warnings on not overpowering the device's input by accident and to do that one was advised to get a set of fixed attenuators.
For the reference: Absolute max. input level = +6dBm with 0dB internal attenuation

I suppose the starter set consisting of 3dB, 6dB, 10dB, 20dB is good enough for the purpose?
I'm talking about these 2W ones:
https://aliexpress.com/item/1005003328270602.html

Any other dB values line up that might be more practical? I'd like to keep the number of units in the set down to 4 at the moment. Maybe some 3dB, 6dB, 15dB, 30dB ? However I see that the 15 dB one is less used than 10dB and 20 dB for some reason.

By the way, I'm mostly going to experiment with the receivers at the moment (building and experimenting with the classic circuits like Direct-Conversion/Superhet/etc.).
Am I overstressing the importance of the attenuators usage for that purpose?
Are attenuators more important for the transmission side when used to protect the SA?

[tszaboo]

I've some of these ones:
www.aliexpress.com/item/4000412303682.html
They measured all right on a VNA up to 4GHz as I remember, don't have numbers now and didn't take too much time to investigate them.

If you test high power stuff, you need high power attenuators. Personally, I would only use these up to +20dBm.

[fourfathom]

It doesn't matter for milliwatt-level receiver testing, but those "2W" attenuators will probably burn out at 2W, and get uncomfortably hot at 1W.  This isn't a problem as long as you know about it.
You use attenuators for several purposes, including (as you know) to reduce the power input to delicate test equipment such as the SA.  But even at low power levels attenuators can improve your measurements by reducing impedance mismatch in a controlled manner. 
For example, if you are testing a filter, it will probably not present a constant 50 Ohm impedance at the input and output ports (especially outside of the filter design frequency), and so the length of coax cable between the signal source, the filter, and the analyzer will also affect the measured filter behavior.  By putting (say) 10 dB attenuators at the filter input and output you can greatly reduce this effect and get a better idea of the actual filter characteristics.
Or, some signal sources do not have a well-controlled output impedance, and so connecting cable length will affect the output amplitude/phase.  Putting an attenuator at the source output will again reduce this effect.

So, I recommend adding a few 6, 10, 20 dB attenuators to your kit.  Probably a few dummy loads as well, although a 20 dB attenuator can often serve this function.

But please be very suspicious of the specs on any of these aliexpress or Amazon attenuators and dummy loads.  In most cases I've found the attenuation and frequency response to be acceptable up to 1 GHz (I generally work in the HF region), but the power ratings are usually wildly exaggerated.

[VSV_electron]

I've some of these ones:
www.aliexpress.com/item/4000412303682.html
They measured all right on a VNA up to 4GHz as I remember, don't have numbers now and didn't take too much time to investigate them.

If you test high power stuff, you need high power attenuators. Personally, I would only use these up to +20dBm.

The ones you referenced are stainless steel while those that are in my original post are copper. Does that matter?
I see people buy both types and are quite happy with either.

[CaptDon]

15dB is kind of uncommon due to the odd math. 3dB = 2X  6dB = 4X  10dB = 10X  and 20dB = 100X when speaking of power (not voltage)

[tszaboo]

I've some of these ones:
www.aliexpress.com/item/4000412303682.html
They measured all right on a VNA up to 4GHz as I remember, don't have numbers now and didn't take too much time to investigate them.

If you test high power stuff, you need high power attenuators. Personally, I would only use these up to +20dBm.

The ones you referenced are stainless steel while those that are in my original post are copper. Does that matter?
I see people buy both types and are quite happy with either.

Steel is more durable, and generally are machined to better tolerances.
And if it's not, it can totally wreck your brass SMA connector, so don't try to place it directly on your most expensive equipment without trying first.
If you are already shopping for these, throw in a DC block, and a couple Dummy load to your cart.

[VSV_electron]

...
If you are already shopping for these, throw in a DC block, and a couple Dummy load to your cart.

What are those parts, could you please elaborate a little on them because I feel if I search just for "DC block" it will give me too broad a choice of variants in a a variety of contexts, not necessarily related to RF.

Quick edit: I had a look on Ali Express and they have something like this:
https://aliexpress.com/item/1005004188972062.html
It's apparently a dummy load rated at 5 W. What do I need it for in the context of my beginner RF experiments with the TinySA ?

[tszaboo]

...
If you are already shopping for these, throw in a DC block, and a couple Dummy load to your cart.

What are those parts, could you please elaborate a little on them because I feel if I search just for "DC block" it will give me too broad a choice of variants in a a variety of contexts, not necessarily related to RF.

Quick edit: I had a look on Ali Express and they have something like this:
https://aliexpress.com/item/1005004188972062.html
It's apparently a dummy load rated at 5 W. What do I need it for in the context of my beginner RF experiments with the TinySA ?

DC block is practically a high pass filter. Some radios output DC voltage on their output, so if you connect it to other equipment if can be problematic. For example a 0dBm output amplifier, you expect 1mW amount of power. But it can be 0dBm radio signal on top of (for example) 2.5V, that supposed to power another amplifier, which is a lot more power into a 50 Ohm resistor.
A DC load is basically a 50 Ohm resistor. You want to have this on any unused outputs, when testing things.

[VSV_electron]

DC block is practically a high pass filter. Some radios output DC voltage on their output, so if you connect it to other equipment if can be problematic. For example a 0dBm output amplifier, you expect 1mW amount of power. But it can be 0dBm radio signal on top of (for example) 2.5V, that supposed to power another amplifier, which is a lot more power into a 50 Ohm resistor.
A DC load is basically a 50 Ohm resistor. You want to have this on any unused outputs, when testing things.

Regarding the DC blocks, yes - I spent some time reading the information on feeding any RF signals even at the pro labs to the input of their SAs through DC blocks. Some say, as you of course already clarified it, that the DC block is simply an HP filter and basically is just a capacitor in series.

Looking at the "not really bargain" prices for even the budget DC blocks like these: https://aliexpress.com/item/1005002890619143.html
... I wonder - are those prices justified for a metal casing with the series capacitor inside? The things must be not as simple as I'm trying to downgrade them of course.
Anyway since you do not object against those budget Asian attenuators I assume the above DC blocks are equally acceptable up to a few hundred Mhz?

As to the dummy loads, it's hard for me to imagine at the moment the amount of unused outputs in my prospective circuits so I'm not sure what number of dummy load units I need to get. To be honest it's even hard for me to imagine any unused inputs at all though that's obviously just the lack of my experience and fundamental knowledge.
Say, I'm experimenting with some kind of a mixer and that should have only one output to the next stage of the RF circuitry and that output goes to the SA input of course. Never mind... I can start with just one dummy load unit and see what comes up next.

[tszaboo]

DC block is practically a high pass filter. Some radios output DC voltage on their output, so if you connect it to other equipment if can be problematic. For example a 0dBm output amplifier, you expect 1mW amount of power. But it can be 0dBm radio signal on top of (for example) 2.5V, that supposed to power another amplifier, which is a lot more power into a 50 Ohm resistor.
A DC load is basically a 50 Ohm resistor. You want to have this on any unused outputs, when testing things.

Regarding the DC blocks, yes - I spent some time reading the information on feeding any RF signals even at the pro labs to the input of their SAs through DC blocks. Some say, as you of course already clarified it, that the DC block is simply an HP filter and basically is just a capacitor in series.

Looking at the "not really bargain" prices for even the budget DC blocks like these: https://aliexpress.com/item/1005002890619143.html
... I wonder - are those prices justified for a metal casing with the series capacitor inside? The things must be not as simple as I'm trying to downgrade them of course.
Anyway since you do not object against those budget Asian attenuators I assume the above DC blocks are equally acceptable up to a few hundred Mhz?

I'm getting a Mini-Circuits DC block and one from Aliexpress in the next few weeks and planning to compare them.
But yes, the DC block is "just a capacitor" the same way, as an attenuator is just some resistors and an Intel i7 is just some transistors on a die. But unlike the i7 it's something you can build yourself.
You don't need a large dummy load, unless it's to test the output of radios.

[radiolistener]

Any other dB values line up that might be more practical? I'd like to keep the number of units in the set down to 4 at the moment. Maybe some 3dB, 6dB, 15dB, 30dB ? However I see that the 15 dB one is less used than 10dB and 20 dB for some reason.

Most usable values are 10 dB, 20 dB and 40 dB, sometimes 6 dB is also needed.

You can combine them, for example:
16 dB = 6 dB + 10 dB
30 dB = 10 dB + 20 dB
50 dB = 40 dB + 10 dB
60 dB = 40 dB + 20 dB
70 dB = 40 dB + 20 dB + 10 dB

20 dB and 40 dB are most usable, because this is technically 1:10 and 1:100 for voltage.

But be careful, Chinese attenuators are single direction, it means that it has input and output. High power input has "SU" mark (usually male). If you put high power on output connector you can burn attenuator.

[Solder_Junkie]

Forget the 3 and 6 dB attenuators for use with a spectrum analyser, it just complicates what you see on the display. 10 dB and 20 dB are all you need for low power attenuators.

If you are testing transmitters, you need power attenuators and these can be expensive. You can make yourself a 40 dB resistive tap that will handle 100 Watts from HF through around 500 to 700 MHz for little cost. This article shows the construction:
http://on4khg.be/wordpress/wp-content/uploads/2015/02/Simple-RF-Power-measurement-W7ZOI-W7PUA.pdf

[Wallace Gasiewicz]

If you are not concerned about absolute value and are worried about your SA input, you can use a "sniffer" ..... just a little antenna.  I use an oscilloscope probe with the GND connection clipped to the tip. You can put it near the circuit to be  tuned and see if the freq is correct and then tune for Max or Min as desired.    I think there are several threads about RF sniffers on this blog.
A DC block is a really good idea, lots of radio circuits have both DC and RF on them.     

Also attaching a piece of TE will change the characteristics of a radio circuit, Using a sniffer will change the circuit much less. Just putting your finger near some circuits will change their characteristics. Probing a crystal oscillator may even shut off the oscillation.     

Edit: for testing transmission, you need a big attenuator. More commonly repair tecks use a big 50 Ohm LOAD.   
The attenuator I use is a 40 dB 100 Watt. Mostly I use a LOAD. For transmission, usually the bigger the Watt ratting, the better.     
A cheap safe way to use your SA on a transmitter output is this:   
Get an appropriate RF LOAD, again bigger is better.   Connect to your transmitter with COAX.     Wrap about 6 turns of wire around the outside of the coax and attach your SA to the wire.  This is usually a lot less expensive but will not give you absolute values.  But it will give you a look at the output spectrum. Use a RF Wattmeter for power readings..    There are a lot more big RF loads for sale used than big attenuators and they are less expensive.
Some CHEAP "Cantenna" Loads had a small circuit to pick off a small part of the transmit signal for use with a Scope or SA. A Cantenna is just a big resistor in a paint can full of oil for cooling.  They usually work in HF OK but not much higher.     

Many of the circuits you will be measuring in radios do not lend themselves to "termination" in 50 Ohms, they will have all sorts of impedance values.  In most radios only the input (antenna) and the output of a transmitter are 50 Ohms. If you have some idea of what the output impedance of your mixer is you can just use the same value of small resistor to terminate it. It's impedance is likely many times higher than a 50 ohm load.

[David Hess]

15dB is kind of uncommon due to the odd math. 3dB = 2X  6dB = 4X  10dB = 10X  and 20dB = 100X when speaking of power (not voltage)

Oscilloscope calibration includes 14dB for 5x voltage, which might be made with 6dB and 8dB.

         Power  Voltage
 3dB       2    1.41
 6dB       4     2
 8dB     6.3    2.5
10dB     10    3.16
14dB     25    5
20dB   100   10

[tszaboo]

Forget the 3 and 6 dB attenuators for use with a spectrum analyser, it just complicates what you see on the display. 10 dB and 20 dB are all you need for low power attenuators.

If you are testing transmitters, you need power attenuators and these can be expensive. You can make yourself a 40 dB resistive tap that will handle 100 Watts from HF through around 500 to 700 MHz for little cost. This article shows the construction:
http://on4khg.be/wordpress/wp-content/uploads/2015/02/Simple-RF-Power-measurement-W7ZOI-W7PUA.pdf

I guess all the SA will have an "External attenuation" setting. Otherwise your statement is true, 10 and 20 usually enough to get the signal to reasonable level. Like a TinySA will take +6dBm max, so 26dBm with the attenuator, which is 300mW. Definitely NOT OK for 36dBm, which is 4 watt.

[Solder_Junkie]

No, the spectrum analysers that I have, and those I used in industry, didn’t have a compensation feature to correct for external attenuation.

If you stick to 10 dB and multiples of 10 dB, it is easy to measure levels in dBm.

Try and work with dBm and 50 Ohms, it gets messy trying to figure out Voltages shown on an oscilloscope. Many (most) modern RF circuits are based on 50 Ohm sources and terminations as it makes design and testing much easier.

If you check out my home made modular transmitters and receiver, you will see what I mean:
https://www.qsl.net/g4aon/

SJ

[tszaboo]

No, the spectrum analysers that I have, and those I used in industry, didn’t have a compensation feature to correct for external attenuation.

If you stick to 10 dB and multiples of 10 dB, it is easy to measure levels in dBm.

Try and work with dBm and 50 Ohms, it gets messy trying to figure out Voltages shown on an oscilloscope. Many (most) modern RF circuits are based on 50 Ohm sources and terminations as it makes design and testing much easier.

If you check out my home made modular transmitters and receiver, you will see what I mean:
https://www.qsl.net/g4aon/

SJ

Anritsu site master:

Reference Level: The reference level is the top graticule line on the display,
and can be set from +30 dBm to –130 dBm. A value may be entered from the
keypad, use the ± key for a minus sign. After entering the value press
the dBm submenu key or the Enter key. The Up/Down arrow keys change
the reference level in 10 dB steps, and the Left/Right arrow keys change the
value by 1 dB. The rotary knob changes the value by 0.1 dB per click. The
reference level value may be modified by the reference level offset value to
compensate for an external attenuator.

R&S:

You can define a reference offset for the reference level. With a reference offset, you
can increase the reference level by a certain amount. This is useful, for example, if an
attenuator or amplifier has been inserted before the RF input. The R&S FSC
automatically takes the loss or gain into account when the level is displayed and no
manual calculations are necessary. A loss introduced at the RF input must be entered
as a positive number and a gain as a negative number.

I'm sure others will do this as well.

[Solder_Junkie]

Have you checked the price of those instruments? 

[tszaboo]

Have you checked the price of those instruments? 

I know the approximate pricing . Also rented them a few times when it was necessary.
The site masters are not very expensive anymore used BTW, less than a used car.
They also don't need a 2W attenuator. The TinySA that I have at home has the setting for the external attenuator.

[ftg]

For what it's worth the Siglent SSA3032X also has reference level offset.
I'd expect it to be a fairly basic feature in a modern spectrum analyzer.