New users guide to REW

Hopefully this helps some new users, pm me with suggestions etc to make it better.

I’m sure there’s a ton of guides on the internet about how to use REW and other similar sound measurement software, but I thought as I have recently been telling people to get off their ass and use it – I should lend a hand and provide a basic guide to getting started. If you’re not that technically inclined then I know measurement software can be daunting. I like REW because it has 95% or more of the features you’d get in professional paid for software such as Systune or Smaart, but it is free. Although if you get enough use out of it I strongly encourage making a donation to the author to allow his work to continue. It is invaluable for building and prototyping enclosures, electronics, crossovers, etc. Hopefully this will give you some ideas into what can be achieved. If you can get results after using this guide then other software should be easier to approach too, they all fundamentally work the same way.

Some people dismiss measurements as being inferior to the ears. I don’t think that debate is necessary here but my opinion is that once you understand how make and interpret these measurements correctly (whether a single piece of equipment or a whole system) then you will get the best of both worlds. But you can’t rely on a measurement unless you are sure that you have made it right, and I think this is where a lot of people get put off. Ears are easier to use, that is a fact! Deeply interpreting measurements made does require some miore specialist knowledge – complex numbers and control theory help a lot. But everyone can look at an SPL graph and figure out how that relates to sound. Once you get that then you can start thinking about phase and then go forwards from there.

This guide will cover setting up your sound interface, and making your first measurements with REW. I will focus on SPL and phase for now, once you have a measurement you can investigate the rest of the analysis options at your leisure by clicking all the main buttons. When people think of measurement software like REW they generally tend to think that a measurement microphone is required – not necessarily so. In fact I will start off by showing you how to measure the frequency response of electronic processing equipment such as a mixer, or crossover (also perfect for figuring out what that mystery black box preamp is really doing when you turn its controls or determing its crossover frequencies). REW can also measure impedance across the bandwidth of audio, so with that you can test the impedance of drivers in enclosures, drivers in free air, crossover networks, individual components such as capacitors/inductors etc. The method for that is slightly different to a basic audio measurement and so perhaps I will cover that another day.

When you test a driver in an enclosure you generate a soundfile – usually a logarithmic sweep of frequency with a known constant amplitude and phase. This is played through one of your soundcard channels, hits your processing gear and amplifiers, is reproduced by a speaker, converted by a measurement microphone back to an electronic signal, and then measured by an input channel of the same soundcard. The difference between what goes out of the soundcard and what enters it again, is the transfer function of the system the signal travels through. You only need one soundcard channel for an REW measurement because there is only one measurement microphone. The other channel is often used as a timing reference which improves measurement accuracy. The measurement you make here is the response of every bit of equipment in that signal chain acting together to produce the transfer function already mentioned. Gains multiply, and phases add/subtract to produce the end result (which is then usually made much weirder by room interactions with real sound waves if you are measuring a speaker, but that is another story).

We can simplify this process for the learning phase by considering a single piece of equipment on its own so that what we measure is solely attributable to it only. We can also skip the microphone which lowers the standards of soundcard required. (Proper measurement microphones require phantom power and therefore pricier soundcards). To make an electronic test we can simply plug the soundcard output to the input of equipment channel 1, and the output of equipment channel 1 to the input of the soundcard. In this way we measure the effect of the equipment on the sound we pass through it. For this guide I will be using a dbx 223XL crossover and an external soundcard with phono inputs/outputs. You could use an internal soundcard, although the noise floor will probably be higher. You will need a way of seperating the left and right signals fro m the soundcard, so if using a card with 3.5mm stereo jacks you will need to be able to break them out into mono signals using an adaptor cable suitable for pluggi.

Another reason for learning this way is that the audio sweeps used to test actual speakers can really piss people off depending on where you live! So keeping it electronic only makes the learning phase a little more preferable for your family/neighbours/pets.

Remember that what REW measures is known as a transfer function. If we know the transfer function of a piece of equipment then for any frequency of sine wave that we input, we know how the output wave will be affected in terms of amplitude in dB (SPL) and phase in degrees (or radians if you prefer). When we measure a piece of equipment that is supposed to pass a signal without altering it, the SPL measurement will be flat and the phase should be 0. When we measure a crossover or filter we will see how the filter shapes the sound in the SPL curve and the phase will also change – it always must with analog equipment if amplitude does as a function of frequency.

So to start, download REW (http://www.roomeqwizard.com/), install it, and fire it up. You should see the screen below after selecting the top right hand preferences icon. I have Windows installed but the guide should be valid for any OS.

 

Figure 1 – Select source

In the preferences window that opens select the input and output channels of the soundcard that you will be using (red circles). Set all the other options as they are shown (they should default to these settings on a new install). Select the analysis tab (figure 2) and make sure that use loopback as a timing reference is enabled. The right hand channel of our soundcard is going to be used for this.

What we are going to do is perform a loopback check using the soundcard input and output you have chosen. So get some cables and directly connect the left input and output together, and the right input and output together. Now select the calibrate button (blue circle – figure 1) from the soundcard tab of the preferences window. A loopback measurement allows us to determine the frequency response of our soundcard so that we can subtract it from future measurements of other equipment to get their true transfer functions. It is also a great test to make sure we have everything connected right. If we can’t get this working, we’re not going to be measuring anything else correctly.

REW is quite good at giving instructions while you calibrate the loopback. Read what it tells you in the prefernces window text box and then click next. Remember we are using the right channel as a timing reference, and the left as the signal.

 

Figure 2 – Select source

Follow the instructions REW gives you regarding your system volume settings. If the preference window options are grayed out, follow them by adjusting your OS volume settings. Leave sweep level at -10dbFS. When ready click next. The soundcard will begin to output a 1KHz signal and you should see both the input and reference channel receiving a signal too. If not, check your wiring. Use the input level control of REW or your OS settings and adjust such that the input channels are reading about -12dbFS (see figure 3). I find this works well as a balance between headroom before ckipping whilst providing a good signal to noise ratio. Very important: If you have other programs installed or OS settings that alter the frequency output of your soundcard such as bass boost etc – turn them off/disable them now. Or you will get weird results.

Press next to accept the setting when you are happy, and then press next once more to start the calibration measurement. The sweep will progress and when it is finished you will be shown the results – see figure 4.

What we expect to see for the soundcard calibration is a flat response of 0dB (i.e. level out=level in) for nearly all of the audio passband. Beneath about 20Hz the response will begin to roll off due to high pass filters designed to block DC in the soundcard and it will also roll off sometime after 20KHz to prevent aliasing from the DAC in the soundcard travelling further down the signal chain. Notice this is a much steeper roll off than the ~20Hz high pass filter. If you have a better, higher bandwidth soundcard than me your high frequency roll off may start above 20KHz. The ripples in response near the high pass roll off are a result of the DACs anti aliasing filter. They look bad but if you look closely then they are <0.25db in amplitude – not a big deal. Again if you have a better soundcard then me it is probably going to look nicer here. But this is exactly the sort of effect that want to remove from the measurements of other equipment by performing calibration. If your response doesn’t look at all like this or has weird bumps in its response, check your wiring and check there is no extra EQ enabled in your soundcard drivers or OS settings and resolve this problem before continuing.

 

Figure 3 – Adjusting input level volume for calibration – good input signal on both channels

 

Figure 4 – Soundcard calibration results

When calibration is finished then you should be returned to the original preferences window. Before you leave it make sure to click the ‘make cal’ button (figure 1, green circle) and select the file you have just made as your calibration measurment. If you do not then it will not be applied to your subsequent measurements

Now the fun part – real measurements. Find whatever bit of gear you wish to measure and connect it up. Connect the left sound card output to left input of the equipment, and the left output of the equipment to the left input of the soundcard to make a loop. Leave the right soundcard output connected to the right soundcard input. We can only test one channel of the equipment at once at this is our timing reference.

Set the equipment to a flat state. Set all gains to 0db, faders to a known position, EQ to flat etc. The measurement that we make is not going to be calibrated well in terms of an absolute dB level. But what we can do is display all measurements after the first in relative terms to it. So the first measurement we make should be as flat and neutral as possible – it will be our reference. Everything else after will be expressed in terms of louder or quieter than this first reference. Remember that REW thinks we are using a microphone, so we have to fool it it here a little and hence why the absolute dB scale is not going to be of use to us (it isn’t useful with a mic either unless you have an SPL meter to calibrate it with). There is a method of fudging the first measurement to be close to 0db which I wil explain later but it isn’t perfect. However generally we are only interested in relative measurements of SPL (unless you are the noise police or are trying to boast about your new cabs). And phase measurement is unaffected entirely. So we can live with this. What we want to do is measure the equipment ‘flat’ and then see how the response changes either. E.g. if I put the low pass channel gain of a crossover up by 6db, I expect to see my measured curve move up with 6dB relative to my original measurement.

When you are happy with the settings of the equipment then click the measure button (Figure 5, purple circle). You’ll get a dialog box where you can choose the bandwidth of the measurement and how long it will take. Leave it on the defaults for now. Click ok, and the measurement will begin. When it is finished you will see a screen like that shown in figure 5 and the measurement details are shown in the orange circle where you can change the filename, save the measurement, and delete it. As you make more measurements they will be added below – in this example I have made 5 already. After the first measurement if you are not happy with the db scale and would like it to be closer to zero click the SPL meter button. In the SPL calibration box enter the value your measurement came in at, but inverted. E.g. if my first reference measurement was at +9db in the passband of interest I would enter -9 in the calibration box. Take the measurement again, your passband should now be a lot closer to zero. Remember that this really doesn’t matter, and is solely personal taste. We care about the relative differences between measurements as we change settings. The measurement becomes noisy at about -70db which is the noise threshold of my soundcard. If yours is better then you should get a less noisy signal than I did.

 

Figure 5 – Make the first measurement

The measurement shown in figure 5 is of a crossover high pass channel set to 3.5KHz. Display the phase if you like by clicking the tick box in the red circle. The measurement looks like it is expected to, we have a flat passband from about 6KHz upwards and below that the SPL response decreases. If we were to measure the low pass signal too, we would expect the SPL curves to cross at (or very close to) 3.5KHz.

One measurement alone is not very useful, so lets make some more. Change a setting (gain, frequency etc) and then measure the unit again. Once you have another measurement select the overlay icon (figure 5, black circle) and then select the SPL and phase curves you wish to compare. I measured the high pass of my crossover with +6dB, then switched to the low pass output which I measured flat with 3.5KHz crossover, flat with 40Hz high pass filter enabled, and +6db. All of these measurements are shown together in figure 6.

 

Figure 6 – Compare measurements

This series of measurements doesn’t tell me a whole lot (apart that the high pass gain of my crossover is a tiny bit more then the low pass) but it is hopefully a good way to demonstrate the potential of the (very cheap!) technique.

Some ideas for using REW:

If you have kit with unknown crossover points or EQs you can figure them out

You can check for general weirdness and anomalies in responses

If you think some system setting is better than another via your ears – find out exactly why

Measure real speakers and align phase between their whole bandwidths. See if it agrees with your ears

Investigate the advanced analysis options, try impulse responses, harmonic distortion measurements, room decay vs frequency etc

See how low your scoops really go :) (Sorry, couldn’t resist).

Have fun! PM me if you want this document with images in higher resolution

 

 

 

 

 

 

 

 

 

 

 

Edited by odc04r - 08 October 2015 at 7:07pm

I see we're not the only people using this bit of kit. It's a very handy tool.

+1

I had a play with with REW a while ago, but found the RTA a bit strange as, iirc, it starts off with the bandwidth from 25hz - 250Hz, or something like that. Still, a useful tool for starting out.

Hi Bee, I'm afraid I haven't. To my shame I haven't tried ARTA/LEAP yet either although they are very much on the todo list. Although REW seems aimed more at the domestic HiFi/HT end of the market I think it is absolutely more than sufficient for an amateur to semi-pro user or cab builder to get measurements that will help improve their setup or designs. Then if you like what you see, maybe it is worth getting into the heavier stuff. Although REW should not be discounted as lightweight, there are some powerful processing options if you want to get into them.

I totally agree with using a decent soundcard. I usually use an ART Pre for 'real' measurements but I wanted this guide to aim at the people who might get put off trying measurements such as this because they might think that without a decent soundcard there is no point in bothering. And I also wanted to point out that you don't strictly need a measurement microphone to practise and get your first results in. It's all about using the right tool for the job (as always and ever). People should be using this type of measurement to supplement their ears, after all you will one day become more deaf, a microphone will not.

Edit: I should add to the guide that I used a fairly cheap external USB soundblaster card for this, quite old, no frills. If you use an internal soundcard expect the measurements to be much noisier. You may hit a noise floor at -40 to -50db (mine was about -70db). A decent high bandwidth external soundcard should be able to get from -80db to -100db. Is this necessary? That depends entirely on the measurements that the user wishes to make of course. If people care about -100db electronic measurements then they probably already own some more sophisticated equipment to do the job


Edited by odc04r - 09 October 2015 at 12:06pm

bee wrote: @ odc04r Have you compared this software to the likes of smaart and clio, Both smart and clio are £1000 plus software packages, How do you think Rew shapes up for the basic functions, and for a diyer does it cover there needs. Ive not used this one, as I own clio and smaart. My software pre clio and smart was holms response, a very good free software as another option to rew... One thing to mention is yousing the lower end stuff, I would still advise using a good quality soundcard, like the focus right stuff and not the internal soundcard built in to your laptop..... http://www.holmacoustics.com/